1_one_pixel_attack_cifar10 - ritika.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MZBcUzvF-3zA"
      },
      "source": [
        "# One Pixel Attack Tutorial\n",
        "## Part 1 - Cifar10\n",
        "\n",
        "### Dan Kondratyuk\n",
        "### September 15, 2019"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "L5zaZWTc-3zB"
      },
      "source": [
        "![](images/pred2.png \"All it takes is one pixel\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mXnnZzVz-3zC"
      },
      "source": [
        "This notebook will demonstrate the one pixel attack with a few different convolutional neural network models. By using differential evolution, we find a special pixel that can modify a target image such that the network misclassifies the image (which it previously correctly classified).\n",
        "\n",
        "In theory, we want models that don't get fooled by such tiny changes. Especially in images, it is undesirable to have a small alteration in the input result in a drastic change in the output. However, even the most accurate neural networks are susceptible to this type of attack.\n",
        "\n",
        "To read more about it, see [the original paper](https://arxiv.org/abs/1710.08864), or the authors' [official repo](https://github.com/Carina02/One-Pixel-Attack).\n",
        "\n",
        "Let's get started."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "z7jdpaMq-3zC"
      },
      "source": [
        "## Imports"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bC9aGyiQ_B3e",
        "outputId": "bfdd15b3-e043-42e8-caf1-134adfefe5ef"
      },
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "pwd"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "id": "am3CXmVJ_IXE",
        "outputId": "0af6b0e9-bef8-4d57-fb4f-92f82e932fff"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'/content'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {},
          "execution_count": 2
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import os\n",
        "\n",
        "path = '/content/drive/MyDrive/one-pixel-attack-keras-master'\n",
        "\n",
        "os.chdir(path)\n",
        "\n",
        "print(\"Current working Directory: \", os.getcwd())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tMxwkEkU_L_9",
        "outputId": "4563ece3-0504-4b05-ae5a-0a9bdc10ab0a"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Current working Directory:  /content/drive/MyDrive/one-pixel-attack-keras-master\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import pickle\n",
        "import matplotlib\n",
        "from keras.datasets import cifar10\n",
        "from keras import backend as K\n",
        "from resnet import ResNet"
      ],
      "metadata": {
        "id": "b23H6-SsAOAe"
      },
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2el-pozr-3zC"
      },
      "source": [
        "Ensure that you have `numpy`, `pandas`, `scipy`, `matplotlib`, `tensorflow-gpu`, and `keras` installed.\n",
        "\n",
        "Alternatively, you may [run this notebook in Google Colab](https://colab.research.google.com/drive/1Zq1kGP9C7i-70-SXyuEEaqYngtyQZMn7). Note: colab allows you to run this notebook on GPU, free of charge. Simply select \"GPU\" in the Accelerator drop-down in Notebook Settings (either through the Edit menu or the command palette at cmd/ctrl-shift-P)."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1eExufe9-3zD"
      },
      "source": [
        "## Load Dataset"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IGPlqkMs-3zE"
      },
      "source": [
        "For this attack, we will use the [Cifar10 dataset](https://www.cs.toronto.edu/~kriz/cifar.html) packaged by Keras. The task of the dataset is to correctly classify a 32x32 pixel image in 1 of 10 categories (e.g., bird, deer, truck).\n",
        "\n",
        "The code below will load the Cifar10 dataset. Keras will need to download the dataset if it is not cached locally already."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "81ealFQg-3zE",
        "outputId": "2be512c4-c2e2-4fa5-cd77-25f8bdea9810"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz\n",
            "170498071/170498071 [==============================] - 2s 0us/step\n"
          ]
        }
      ],
      "source": [
        "(x_train, y_train), (x_test, y_test) = cifar10.load_data()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "4IfJaAw4-3zE"
      },
      "outputs": [],
      "source": [
        "class_names = ['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck']"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "G7B0CVYR-3zE"
      },
      "source": [
        "We can access and display any image in the dataset by its index. For instance, here is a horse."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "wAl6982u-3zE",
        "outputId": "ea0a88de-ba32-48c2-967b-cfd508f65f24"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "import helper\n",
        "image_id = 99 # Image index in the test set\n",
        "helper.plot_image(x_test[image_id])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eEwm2GdE-3zE"
      },
      "source": [
        "## Image Perturbation"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RthNp558-3zE"
      },
      "source": [
        "To begin, we need a function to modify one or more pixels in an image.\n",
        "\n",
        "We can define the perturbation of a pixel as a 5-tuple\n",
        "\n",
        "$$\\textbf{x} = (x, y, r, g, b)$$\n",
        "\n",
        "where $x, y$ are the coordinates of the pixel from 0 to 31, and $r,g,b$ are the red, green, and blue values from 0 to 255. Then multiple perturbations can simply be a concatenation of these tuples:\n",
        "\n",
        "$$X = (x_1, y_1, r_1, g_1, b_1, x_2, y_2, r_2, g_2, b_2, ...)$$\n",
        "\n",
        "We could instead use an array of tuples, but the optimization algorithm we will use requires it to be a flat 1-d vector.\n",
        "\n",
        "Then the function to perturb an image can take as an input the image and $X$, and output a copy of the image with each pixel at $x_i, y_i$ modified to have the color $r_i, g_i, b_i$. To speed up computation, we will batch together an array of $X$ perturbations, denoted $X_S$."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "id": "-ztE7Hpf-3zE"
      },
      "outputs": [],
      "source": [
        "def perturb_image(xs, img):\n",
        "    # If this function is passed just one perturbation vector,\n",
        "    # pack it in a list to keep the computation the same\n",
        "    if xs.ndim < 2:\n",
        "        xs = np.array([xs])\n",
        "\n",
        "    # Copy the image n == len(xs) times so that we can\n",
        "    # create n new perturbed images\n",
        "    tile = [len(xs)] + [1]*(xs.ndim+1)\n",
        "    imgs = np.tile(img, tile)\n",
        "\n",
        "    # Make sure to floor the members of xs as int types\n",
        "    xs = xs.astype(int)\n",
        "\n",
        "    for x,img in zip(xs, imgs):\n",
        "        # Split x into an array of 5-tuples (perturbation pixels)\n",
        "        # i.e., [[x,y,r,g,b], ...]\n",
        "        pixels = np.split(x, len(x) // 5)\n",
        "        for pixel in pixels:\n",
        "            # At each pixel's x,y position, assign its rgb value\n",
        "            x_pos, y_pos, *rgb = pixel\n",
        "            img[x_pos, y_pos] = rgb\n",
        "\n",
        "    return imgs"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7DmPD1H8-3zF"
      },
      "source": [
        "Now we can modify the pixels of any image we want.\n",
        "\n",
        "Let's modify our horse image by making pixel (16,16) yellow."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "-4ARL5wx-3zF",
        "outputId": "1c0df47f-da3d-4f37-c215-b5213b648906"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "image_id = 99 # Image index in the test set\n",
        "pixel = np.array([16, 16, 255, 255, 0]) # pixel = x,y,r,g,b\n",
        "image_perturbed = perturb_image(pixel, x_test[image_id])[0]\n",
        "\n",
        "helper.plot_image(image_perturbed)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lCv7bQoS-3zF"
      },
      "source": [
        "## Load Models"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "60PLR6nT-3zF"
      },
      "source": [
        "To demonstrate the attack, we need some neural network models trained on the Cifar10 dataset. We will now load some pretrained models, which can be found in the `networks/models` directory.\n",
        "\n",
        "It is recommended to use Keras with a GPU enabled. If you're [running in Google Colab](https://colab.research.google.com/drive/1Zq1kGP9C7i-70-SXyuEEaqYngtyQZMn7), you can enable a GPU instance by selecting `Runtime > Change runtime type > Hardware accelerator > GPU` (you will need to re-run all cells). The code below can be used to check (if using TensorFlow)."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "id": "0tUgkGxQ-3zF"
      },
      "outputs": [],
      "source": [
        "# Should output /device:GPU:0\n",
        "#K.tensorflow_backend._get_available_gpus()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "F7NIm5nr-3zF"
      },
      "source": [
        "There are two models included in this repository, `lenet` and `resnet` which will be loaded from disk automatically.\n",
        "\n",
        "Optionally, you may [download the larger, more accurate models](https://www.dropbox.com/sh/dvatkpjl0sn79kn/AAC9L4puJ_sdFUkDZfr5SFkLa?dl=0) (e.g., Capsule Network, DenseNet, etc.). Make sure to copy the models into the `networks/models/` directory. Then uncomment the lines below and run the cell to load the models of your choosing."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xpynAQb_-3zF",
        "outputId": "c3ca75d6-7dcc-4a0a-998f-30b3bea12d41"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Successfully loaded resnet\n"
          ]
        }
      ],
      "source": [
        "#lenet = LeNet()\n",
        "resnet = ResNet()\n",
        "\n",
        "models = [resnet]\n",
        "\n",
        "## Uncomment below to load more models to play with. Make sure the model files exist by training or downloading them.\n",
        "\n",
        "# lenet = LeNet()\n",
        "# pure_cnn = PureCnn()\n",
        "# net_in_net = NetworkInNetwork()\n",
        "# resnet = ResNet()\n",
        "# densenet = DenseNet()\n",
        "# wide_resnet = WideResNet()\n",
        "# capsnet = CapsNet()\n",
        "\n",
        "# models = [lenet, pure_cnn, net_in_net, resnet, densenet, wide_resnet, capsnet]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hEu1nWbW-3zF"
      },
      "source": [
        "Note that there are even more networks available in the `networks` directory, but must be trained before loading them here."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "y3OarlIp-3zF"
      },
      "source": [
        "### Calculate Model Accuracies"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jIzmdefX-3zF"
      },
      "source": [
        "After loading the models, we would like to evaluate all test images with each model to ensure that we only attack the images which have been classified correctly. The code below will also display the accuracy and number of parameters of each model."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "scrolled": false,
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 124
        },
        "id": "zAIqfgFp-3zF",
        "outputId": "5f0f62b0-184c-4483-f129-3b9d0480dba8"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Evaluating resnet\n",
            "79/79 [==============================] - 51s 638ms/step\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "     name  accuracy  param_count\n",
              "0  resnet    0.9231       470218"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-d5dd66dd-c048-4a59-b087-b6e2ac608edb\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>name</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>param_count</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>resnet</td>\n",
              "      <td>0.9231</td>\n",
              "      <td>470218</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-d5dd66dd-c048-4a59-b087-b6e2ac608edb')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-d5dd66dd-c048-4a59-b087-b6e2ac608edb button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-d5dd66dd-c048-4a59-b087-b6e2ac608edb');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "  <div id=\"id_bdeb5a16-80f1-4217-b635-13bc11b1a890\">\n",
              "    <style>\n",
              "      .colab-df-generate {\n",
              "        background-color: #E8F0FE;\n",
              "        border: none;\n",
              "        border-radius: 50%;\n",
              "        cursor: pointer;\n",
              "        display: none;\n",
              "        fill: #1967D2;\n",
              "        height: 32px;\n",
              "        padding: 0 0 0 0;\n",
              "        width: 32px;\n",
              "      }\n",
              "\n",
              "      .colab-df-generate:hover {\n",
              "        background-color: #E2EBFA;\n",
              "        box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "        fill: #174EA6;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate {\n",
              "        background-color: #3B4455;\n",
              "        fill: #D2E3FC;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate:hover {\n",
              "        background-color: #434B5C;\n",
              "        box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "        filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "        fill: #FFFFFF;\n",
              "      }\n",
              "    </style>\n",
              "    <button class=\"colab-df-generate\" onclick=\"generateWithVariable('network_stats')\"\n",
              "            title=\"Generate code using this dataframe.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M7,19H8.4L18.45,9,17,7.55,7,17.6ZM5,21V16.75L18.45,3.32a2,2,0,0,1,2.83,0l1.4,1.43a1.91,1.91,0,0,1,.58,1.4,1.91,1.91,0,0,1-.58,1.4L9.25,21ZM18.45,9,17,7.55Zm-12,3A5.31,5.31,0,0,0,4.9,8.1,5.31,5.31,0,0,0,1,6.5,5.31,5.31,0,0,0,4.9,4.9,5.31,5.31,0,0,0,6.5,1,5.31,5.31,0,0,0,8.1,4.9,5.31,5.31,0,0,0,12,6.5,5.46,5.46,0,0,0,6.5,12Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "    <script>\n",
              "      (() => {\n",
              "      const buttonEl =\n",
              "        document.querySelector('#id_bdeb5a16-80f1-4217-b635-13bc11b1a890 button.colab-df-generate');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      buttonEl.onclick = () => {\n",
              "        google.colab.notebook.generateWithVariable('network_stats');\n",
              "      }\n",
              "      })();\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "    </div>\n",
              "  </div>\n"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "dataframe",
              "variable_name": "network_stats",
              "summary": "{\n  \"name\": \"network_stats\",\n  \"rows\": 1,\n  \"fields\": [\n    {\n      \"column\": \"name\",\n      \"properties\": {\n        \"dtype\": \"string\",\n        \"num_unique_values\": 1,\n        \"samples\": [\n          \"resnet\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"accuracy\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": null,\n        \"min\": 0.9231,\n        \"max\": 0.9231,\n        \"num_unique_values\": 1,\n        \"samples\": [\n          0.9231\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"param_count\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": null,\n        \"min\": 470218,\n        \"max\": 470218,\n        \"num_unique_values\": 1,\n        \"samples\": [\n          470218\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    }\n  ]\n}"
            }
          },
          "metadata": {},
          "execution_count": 19
        }
      ],
      "source": [
        "network_stats, correct_imgs = helper.evaluate_models(models, x_test, y_test)\n",
        "correct_imgs = pd.DataFrame(correct_imgs, columns=['name', 'img', 'label', 'confidence', 'pred'])\n",
        "network_stats = pd.DataFrame(network_stats, columns=['name', 'accuracy', 'param_count'])\n",
        "\n",
        "network_stats"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "V1J55zyk-3zF"
      },
      "source": [
        "### Prediction Function"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mpILObZO-3zF"
      },
      "source": [
        "For the black-box attack, all we should care about is the inputs to the model (the images), and the outputs of the model (the prediction probabilities). No special information about the model is required; we could even swap it with a model that is not a neural network.\n",
        "\n",
        "Define a function that runs several perturbed images on a given model and returns the model's confidence (probability output) in the target class, one confidence value per image. If the target class is the correct class, this will be the function that we want to minimize so that the model will be most confident in another class (which is incorrect). Otherwise, the target is an incorrect class and we will want to maximize it."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "id": "1hdSBWeO-3zF"
      },
      "outputs": [],
      "source": [
        "def predict_classes(xs, img, target_class, model, minimize=True):\n",
        "    # Perturb the image with the given pixel(s) x and get the prediction of the model\n",
        "    imgs_perturbed = perturb_image(xs, img)\n",
        "    predictions = model.predict(imgs_perturbed)[:,target_class]\n",
        "    # This function should always be minimized, so return its complement if needed\n",
        "    return predictions if minimize else 1 - predictions"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8vlw_xko-3zF"
      },
      "source": [
        "Below we can modify a pixel in an image and see how the confidence of the model changes. In almost all cases, the confidence will not change. However, for very special cases it will change drastically."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 26,
      "metadata": {
        "scrolled": true,
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 493
        },
        "id": "suUydszA-3zG",
        "outputId": "b9df6587-437e-471f-ea59-3180288475da"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "bird\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Confidence in true class bird is 0.00018887463\n",
            "Prior confidence was 0.70661694\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "image_id = 384\n",
        "pixel = np.array([16, 13,  25, 48, 156])\n",
        "model = resnet\n",
        "\n",
        "true_class = y_test[image_id, 0]\n",
        "print(class_names[true_class])\n",
        "prior_confidence = model.predict_one(x_test[image_id])[true_class]\n",
        "confidence = predict_classes(pixel, x_test[image_id], true_class, model)[0]\n",
        "\n",
        "print('Confidence in true class', class_names[true_class], 'is', confidence)\n",
        "print('Prior confidence was', prior_confidence)\n",
        "helper.plot_image(perturb_image(pixel, x_test[image_id])[0])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gPpFHiWu-3zG"
      },
      "source": [
        "## The Attack"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1OBEi6Pz-3zG"
      },
      "source": [
        "Here we will demonstrate two variants of the one pixel attack: untargeted and targeted."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "81YwzRkE-3zG"
      },
      "source": [
        "### Targeted vs. Untargeted Attacks"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OJQXNOxY-3zG"
      },
      "source": [
        "The objective of an untargeted attack is to cause a model to misclassify an image. This means we want to perturb an image as to minimize the confidence probability of the correct classification category and maximize the sum of the probabilities of all other categories.\n",
        "\n",
        "The objective of a targeted attack is to cause a model to classify an image as a given  target class. We want to perturb an image as to maximize the probability of a class of our own choosing."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2qZIuyDc-3zG"
      },
      "source": [
        "### Success Criterion"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6xtQ9tit-3zG"
      },
      "source": [
        "Define a function so that whenever a given perturbation is sufficient to fool a model, it returns `True`. This will be called the success criterion. The `targeted_attack` boolean flag will indicate whether success means maximization of the target class or minimization of the correct (target) class."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 32,
      "metadata": {
        "id": "AZohgyel-3zG"
      },
      "outputs": [],
      "source": [
        "def attack_success(x, img, target_class, model, targeted_attack=False, verbose=False):\n",
        "    # Perturb the image with the given pixel(s) and get the prediction of the model\n",
        "    attack_image = perturb_image(x, img)\n",
        "\n",
        "    confidence = model.predict(attack_image)[0]\n",
        "    predicted_class = np.argmax(confidence)\n",
        "\n",
        "    print('Predicted Class:', class_names[predicted_class])\n",
        "    # If the prediction is what we want (misclassification or\n",
        "    # targeted classification), return True\n",
        "    if verbose:\n",
        "        print('Confidence:', confidence[target_class])\n",
        "    if ((targeted_attack and predicted_class == target_class) or\n",
        "        (not targeted_attack and predicted_class != target_class)):\n",
        "        return True\n",
        "    # NOTE: return None otherwise (not False), due to how Scipy handles its callback function"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wG5E7p9c-3zG"
      },
      "source": [
        "Here we demonstrate the usage of the success criterion function. It's nearly identical to `predict_class()` as before, but also decides the success of the attack. For purposes of demonstration we assume an untargeted attack."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 33,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 528
        },
        "id": "qV-h5QcJ-3zO",
        "outputId": "ab3752e5-01c5-45a4-df61-bd1feda95150"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "horse\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: cat\n",
            "Confidence: 0.0746012\n",
            "Prior confidence 0.5005424\n",
            "Attack success: True\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAGFCAYAAAASI+9IAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAZYElEQVR4nO3cW49e91XH8bWf83HO9ozHM46dxEmapk3aHNQEKqqqVYUgQIEiLrjiFfBmeAsIFYQQXLSiUlGpWqVA06ZOfcrJ9njGM2PP8Tkf9t5cuF7qTdr1K32oUb8fiSuWV/7Pfg6/2aD9S/I8zw0AADMr/KYPAAB4fBAKAABHKAAAHKEAAHCEAgDAEQoAAEcoAABcKTKUZZnt7OxYu922JElmfSYAwK9ZnufW6XRsfX3dCoWPvx8IhcLOzo5tbm7+2g4HAPjN2Nraso2NjY/934dCod1um5nZv/zj31uz0Qj9h6sHV0NzZmbvbp2GZ83MfnxYDc+++JmXpd2rG0+FZ7NpKu0uCHdZs74jy6X92lkSezwekp/5Pa30H1BPM7trOMsOg/+v/3cE+ZLM8CJmwqxyvfu9rv3VH77mv+cfJxQKj36gmo2GNZvN0AFqg1pozsysXhuHZ83MqtV4KNQbsfM+0mz94gv28wiFj5smFP6Xw0Yo/N/6bQgF/ze/5LvP/6MZAOAIBQCAIxQAAI5QAAA4QgEA4AgFAIAjFAAAjlAAADhCAQDgQk80P9IuZ9Yqx563W21Ow3v/+3BHOYatrH4uPNuoSi/RkmwSnm026tLuPBOeVUy0vFafgFam1Wc3H5enWmf+VLgwK59khk8d54/JE+ePk1k+0Zzl2rs/qyeaJ5VKaI47BQCAIxQAAI5QAAA4QgEA4AgFAIAjFAAAjlAAADhCAQDgCAUAgCMUAABO6oAoJbmVk9jj3ePeaXjv/YFyCrPTTjc8e+PqFWn3659fDc9WCk1pd5an8eHgdfZxsUdhpjUXymH0fgHhINrmWZZiyC9T+XNNXP641JDoZRvqyYX/glBbIW62PFGKK8xS5XUKBykFf1O4UwAAOEIBAOAIBQCAIxQAAI5QAAA4QgEA4AgFAIAjFAAAjlAAADhCAQDgCAUAgJO6j2pJarVCrL9nOo33/Jym0jHs7s698OxLz16SdrebrfBsoSittlIx/g+k/iD7VfpslH4VtVwnvnu2PTza9lwtkFI8RsVKyWPSfqR3H81OLp4mV74Ts+xVyuJ/109Ksd8f7hQAAI5QAAA4QgEA4AgFAIAjFAAAjlAAADhCAQDgCAUAgCMUAACOUAAAOKlf4t7hsXWG49Dsk/Mr4b0XziqnMLtxoxuePXvuCWl3tVYLz2bpVNpdDD5m/pD2aHySaPleyONVB3kmrTZL4mdPhNmf/QthUrsmas2FcnL1ZUonEc+tvJ1q/YP0OtW2DbVuZYZ1HtJ1UWsuhPFM+Ekpl2LXgzsFAIAjFAAAjlAAADhCAQDgCAUAgCMUAACOUAAAOEIBAOAIBQCAIxQAAI5QAAA4qfvo6//2tlUr1dDs115eDu99+dKicgy7eRwv/FhZ0YqVov0gZlqHjJlZIYmfW6yzMbnnReknKs6uQ0YnnGXW3TqZMC+eRTqJWKxUlA4j/t0onUXtMlLN8rOi7J7d61Q6mCrBnx/uFAAAjlAAADhCAQDgCAUAgCMUAACOUAAAOEIBAOAIBQCAIxQAAI5QAAA4qebi5kcTK5Vjj3f/3f0r4b1//ScvKcewVy/EqjbMzFo1abUVk3hOqvUPhSRejJHoz91r40rNhdq5Mev2gllR6wiUP6keo0uoHUWbVmoXZvwJt0QqotG250rNRT67v70zoZolegruFAAAjlAAADhCAQDgCAUAgCMUAACOUAAAOEIBAOAIBQCAIxQAAI5QAAA4QgEA4KTuozwvWJ4VQ7NvfzgI7137j3hPkpnZH33xlfDsJD2WdidSP5HWl6LNiw01Yj+RMq121Cj/QOnKMfsVepi07bNbLXTUmJl4lFm2GanXZHbXUP6szPBTLp1EfO9zZV7pPgreAnCnAABwhAIAwBEKAABHKAAAHKEAAHCEAgDAEQoAAEcoAAAcoQAAcIQCAMBJNRfTdGqWxHIkqdTCe//5uzeUY9h8uxGevfC89BJttxd/bPzy5eek3dFrZ2amtjlEH2H/uf/CDCZ1WhWBSYeZ5W5VIrz3qiyLV7OoZvneyx9ysSlEMdvXqdZzCFJhb3AxdwoAAEcoAAAcoQAAcIQCAMARCgAARygAAByhAABwhAIAwBEKAABHKAAAHKEAAHBSMVCtVrVSuRqarddXw3s/PNhSjmH/8O13wrNP3jmUdhca74Zn//hP/1za/enPfDY8m6VCqYmZ5WIvTEEoS1Jre5Jc+Adi6UyaTcOzo8FA2j2djKX5yXAYnh32+9LuUjF+DRsLy9Luersdni0Wi9LuXPkgip9ZtSopET7jWaYdpiCcRa1sUl6m0qlVLMU2c6cAAHCEAgDAEQoAAEcoAAAcoQAAcIQCAMARCgAARygAAByhAABwhAIAwBEKAAAndR81mm0rB7uPCkI5yOVnX1SOYel4Nzz7zrUPpd216r3w7MLCorT7+Rfir7Nc0jpnrKA1rEg9Mmkm7e71jsKzB1u3pd17N38cP8f+jrR70u9K84NuvFfr8OhU2p1b/P0vLq5Ju59+7ffCs69/6SvS7nQa/6wk5bK0u6QUDpnZ7q33w7OLZ9el3bVmMzwr9UGZWS60H+VZvCMtz2PvDXcKAABHKAAAHKEAAHCEAgDAEQoAAEcoAAAcoQAAcIQCAMARCgAARygAAJxUczEc9i1NY49Vl0uV8N72vPaY/tLCUnj28HBP2n18Gq86uH3nI2l3vxevOphfWpB2Dzodab67txWevfPOW9Lu/fd/Gp6dDHrS7uIk/v6URtruzsGBNL99EH8/J8HvzSOtcvyr+WB0Rdp9cvUH8eHt96TdaSte/3DxpTek3WPxu3zzRvzsr7/5NWl3UajoqAi/hWZm927fCs/ubt8Jz/b7/dAcdwoAAEcoAAAcoQAAcIQCAMARCgAARygAAByhAABwhAIAwBEKAABHKAAAHKEAAHBS95FZ8rP/icjCW3u9gXSKYvgMZm+89llp97UbN8OzjWZb2v2df/9GePaJlvbWbH3/W9K8dY7Co/1hvOPHzKxaivfCtOcb0u5yMT5750DrPqpkWj/RufX18Gx9Qfus1Oqt8OzmaCjtXju4Hp7d/9e/lXZvzW+EZw/f1Tq1dnfuS/NrX/6z8Ox/vfVNaffyUryvrV1fkHY/EPqMCtP4Z3YwiP3OcqcAAHCEAgDAEQoAAEcoAAAcoQAAcIQCAMARCgAARygAAByhAABwhAIAwEldCqViyUrF2D9JkngVRamQK8ewTm8Snk0nfWn30lI9PPvMxfgj/WZm1775T+HZ++NjafdKJX5NzMxq9Up4tlXS/nZYWqiFZ8+d0eofPtw+Cc+2l5el3Y3qijTfGccrBtIs/n0wMzvpxqsrCvlI2l2aWwjPnpt2pN3jhbPh2eO9fWl3cV37vo3S+Gfl9jvvSruXFi+EZzfW4rNmZvPNeMXJxMbh2aQU+43gTgEA4AgFAIAjFAAAjlAAADhCAQDgCAUAgCMUAACOUAAAOEIBAOAIBQCAIxQAAE7rPrJTK1msM2c8jfd3JJYpxzChVsm6g6K0e9Q7Dc9eTrXOmcVJvJ+ov7sr7R4saB1CK0vz4dkzde39qbbjvUppKd41ZWaWNeJvfjKZSrtHY60naziKv5+Hp9pnZTqMn+VMuyzt3kviuyuNVWn3/iT+d+bt/a60+9Il7bMyPNoKz+a51h3WOYl/P+8Xtb+9a8Xz4dnT7v3w7GAY+wxypwAAcIQCAMARCgAARygAAByhAABwhAIAwBEKAABHKAAAHKEAAHCEAgDASTUXzzzVsmq1Gpp964d74b2VarxywcysKDw2Xq0sSLsH3Vp49tatW9Luc1m8dmGjFbvOjyyM4tfbzOzofrz+Y7J+Vtq9Ul8Oz45HqbR7OhmHZxtifUo5G0rzB8fxSpTTI61Co1mLfzVX2lrFSW8U//7cKS1Ku5fr8cqNUkW73u1mU5pvzTXCs+PDY2n3WPhonfS0Co2zafx3olCK/04Ugh8p7hQAAI5QAAA4QgEA4AgFAIAjFAAAjlAAADhCAQDgCAUAgCMUAACOUAAAOEIBAOCk7qPPvPS8NRr10OyVn74f3tsbxftSzMzWV+NdSUWhG8TMbH8vfkmOT+PdN2Zmy0kenl1sSW+NLStlLGY2PNoNzx6l2u5eUeioScSOp2a8m2o+0/qGTsX+m93TUXj2pKv1/Jyfj3/Ge0n8mpiZ9fJ431QziffwmJnVG/Eepuay9t4PK9rfsHkhvr9R1K5hrboQnp1rxWfNzDq3boZnm2tr4dlisFOJOwUAgCMUAACOUAAAOEIBAOAIBQCAIxQAAI5QAAA4QgEA4AgFAIAjFAAAjlAAADipYOfChSet1Yr12qyvzoX3fnDnSDmGbZ5fDc8+OOxKu3f374Vnz65ckHb3Ld4LY8WetLtmiTRfGMd7mA7uxa+JmdkL62fCs4et+HtpZrZ9IvQZjR5Iu9+7eyjN3zmthGf/Jv22tPtb2VfDs8sW7zIyM1trxee3xhNp94NuvA9qL12Wdr+yel6aT/NBeHZpI/6ZNTPb3or/Zt3rar9ByfGd8OwnNuPXpDQthua4UwAAOEIBAOAIBQCAIxQAAI5QAAA4QgEA4AgFAIAjFAAAjlAAADhCAQDgpJqLZrNlzWYrNNtoxSsdFua1x8ArlXiWvX/zqrR7NIhXHTSan5J2Hx2chGcLi7Hr7IrxWhEzs9o0XqOxvzOWdndO4q9zY31d2n3tQby64Ps78Vkzs6sHWlXIK194PTy789LXpd2lb8Tf/7nBrrT7OIl/7a93y9LuYv+D8Oxg8aK0+8VXvyDNF/J4Jcrh9g1p9+7298KzF595Wtp9726sjsLMLCnWhNnYHHcKAABHKAAAHKEAAHCEAgDAEQoAAEcoAAAcoQAAcIQCAMARCgAARygAAByhAABwUvfR6fGhZZNRaPbk9Di8t1jS+lX29uPdOh/e+kjavbQY7xAqipl6sL0Vnp1uXJJ2FxbiHShmZguFW+HZ+nbsPX9k+95eePbckxel3ZfP1MOzP/nJkbQ7WdZ6mJYunAvPLizGz21mVl1dDs/+6MptabctroZH+5NUWp11h+HZl19ek3ZPOg+k+V4v3qk2GsS7wMzMNtbj78/ZJa3HbDpeCM8qPXDTNDbLnQIAwBEKAABHKAAAHKEAAHCEAgDAEQoAAEcoAAAcoQAAcIQCAMARCgAAJ9Vc/OidK1avxx7X7/Xjj7tbUlWOYQ8Oj8Oz43Ff2r25+Vx49vRBvM7BzOyp5Dg8W+rHZ83MuosXpfnymSw8e2ktXhdgZnbQEd77fryyxMzs7tE4PLvy/Kel3V998w+k+e+89XZ4djicSLufffaZ8Oy1g/vS7vdG8cqN/kT7/qzOxSsdnnz6grS7kR9I84XyNDy7cCZeW2FmdnZ5JTw7HsfPYWa2vDIfnm204te72xuE5rhTAAA4QgEA4AgFAIAjFAAAjlAAADhCAQDgCAUAgCMUAACOUAAAOEIBAOAIBQCAk7qPtn/yA6tWyqHZ881KeO/uUDqG7e3FO4eqpaK0e9I9Dc/u79yWdn/h6XinSX60Le0+Kjek+aW5dnh281y858XMbDKNd/Fs3b4r7R4sXQzPvvmXfyHtfvGTT0vzG+fj1+Xa9VvS7usffBievZs2pd2V4HfYzMyq8a4pM7NCOb5784l1aff88pI03xrH+70KBe3v49yS8GySxGfNzPIs3pPVH3TCs+VirIOJOwUAgCMUAACOUAAAOEIBAOAIBQCAIxQAAI5QAAA4QgEA4AgFAIAjFAAATuqXOL/1PauXYjnyQboY3js586pyDOt04lUU02Ff2n376pXw7Plq/HF0M7PzSbyO4LCv1Qvcu35Vmi++8EJ4dmHpjLR7qTcMz7673ZV2v/alN8Kzn3z2krR7Moif28zswtl47cLmqlYV8p9z8dqS965dl3avN2N1B2ZmJbEm5iSrhmdT8W/SbKh9JyaT+OtM1JqLPF6hocqzND6bxl9jdJY7BQCAIxQAAI5QAAA4QgEA4AgFAIAjFAAAjlAAADhCAQDgCAUAgCMUAACOUAAAOKn7qDnoWKOUhGZHk1p47yReIfNw96AXns2mWl+KWbxL5JnF2LV4ZCGPn+VYzGu14+lo/354tr6xJu2uzS/Eh3ttaffm05fjw1OtmyrNtD6bxIT3v5BLu597cjM8e+nShrT7pBvvm1pfiX+PzcyeWoz3ey0uxPudzMwmqfZ+Kh1C4ltvucXfz0R764XNZonwOxGd5U4BAOAIBQCAIxQAAI5QAAA4QgEA4AgFAIAjFAAAjlAAADhCAQDgCAUAgJNqLo4GqQ2LsUf7T8rV8N7xNF4tYWaWZfG6iPklrUOjfLofnl0txR+jNzPLx/HH9EutFWn3fBK/3mZmvc5xeLY/1K5hPoq/P4vn1qXdi8tz8XPkYr+AVDBgludCN4L2UbEsjV/Dckk7d6NRCc+OS1rNRaU9H56tNerS7mFfq3IpKhUQWmONKZ+VTPwcFoU/1dO0qGwOTXGnAABwhAIAwBEKAABHKAAAHKEAAHCEAgDAEQoAAEcoAAAcoQAAcIQCAMARCgAAJ3UfTdJoe4bZpNEM7x2NB8oxLBV6YaoV6SVaOon3MA1SrTAlDfZGmZlVGlrnTKuszU+68R4m4dhmZnbYjXfULH5yQ9rdbDfCs1kmdBM9/BfaeKLMa39/1Vut8Oz5Da0/6kc//mF49vruUNq9vPa74dlOpyvtHg9H0rwl8WveqGndYcp73+lo17BSip97IPSp9fqx31nuFAAAjlAAADhCAQDgCAUAgCMUAACOUAAAOEIBAOAIBQCAIxQAAI5QAAA4QgEA4KRioE5WtnESK8I5ncS7Qe7v3VWOYek03k80Ec5hZjYRdqeJ1pfSrJTDswfdU2n3cKwVFM0trYZny7l2DQd5/HU+cfEJaXdF+DNmLPRYPZRL04VC/DCJ0MNjZlYpxd/Pz73xirS7JOx+TfuJsOeeeyZ+jiTapPaQ2geWC/vzVPuspMJ3oljSruEgje8eCL9v0VnuFAAAjlAAADhCAQDgCAUAgCMUAACOUAAAOEIBAOAIBQCAIxQAAI5QAAA46fnrKx2zcjBGOpVJeG+9otVFlOfnw7OTnlYXUW3Uw7PLDe3x9eFoHB8O1ok8Uq80pPlyPT7/YEerIbHGYnh0beOctDpRKje0S2hZptV5ZNM0PFsua7uno/j3p13Vvj9f/L3Ph2cLJp5bqBaZZPHrZ2bWHwy1s0zj17BW0ipOBuP4ddk76Eq7i0L7R6MS/w0q5rHXyJ0CAMARCgAARygAAByhAABwhAIAwBEKAABHKAAAHKEAAHCEAgDAEQoAAEcoAACcVN7zQXdqxWAnz9qFeD/R5uqacgzrnByGZw8HJ9Lucil+SQaJlqm7o3hfylEa720xM0sa2nyrFC9Yub6v9UfVX/hEePbs6oq0ezSK99/kiVAiY3JVkmVZvOdnPNZ6fvJpfHce7LR5pFiuhGen4udQuYiVstYddv/BSJq/tbMfnv3E5QvS7lojfg2LJ1pnUyZ0Qs3Pt+PnCP62cacAAHCEAgDAEQoAAEcoAAAcoQAAcIQCAMARCgAARygAAByhAABwhAIAwEnPmT/7qRfCNRCVpaXw3sXFOeUYVirGH+s/fRB/1N3MrD85CM9u97V6gfPNanh2MNHqBaojrQJgdBp/nfPrm9LuS6++Gp6tVcrS7vF4HJ4tCp8TM61awsxsMhGueUEr0SgIfRHFkva3XZbH61Yy066hxVebCXUOZmZt4ftjZnZp82x4tlbWKlHKlfj8wlxD2p1Y/Lo0G/FrkuexvdwpAAAcoQAAcIQCAMARCgAARygAAByhAABwhAIAwBEKAABHKAAAHKEAAHCEAgDASd1Hv//ml61er4Vmv/Xdt8N7D09OlGPYypl4p8l8W+tV2plrhWfv3r4p7d5YaoZnE+tIu7V2FbOs3w3PnmkvS7vXl+rxc6RKWY5ZnsY7obJc6zKajrW+qUToM0oK2t9fynSWadfQ0nh/VC70JJmZ5UpXUqL1QTXqWvdRQ+kFEnuY0kn8GpaL2nvfrMR/lgvC5Y7OcqcAAHCEAgDAEQoAAEcoAAAcoQAAcIQCAMARCgAARygAAByhAABwhAIAwEk1F2eW563RiNVclJN4ZUA6GSrHsEYl/nj87Tt3pd2d44Pw7CSpSLvnfucr4dnprevS7ubpR9L8eBx/TL+z90Davf9e/OxLFy5Ku/NJvLqiKFZLWCJ0BpjZRKhGyKfx621mVkyK4dlCSXud0/EoPJskYj2HMK9WfwzG2u+E8m6WS/HrbWamtH/UK9ruRPgcTqfx78N0Gvu8cqcAAHCEAgDAEQoAAEcoAAAcoQAAcIQCAMARCgAARygAAByhAABwhAIAwIVqLvL84WPX/UH8MfPxJF5zMZkIz4yb2WgUrwyYCLUIZmbTNH6WaaadeyiceyqeeyBew7Fw9uFU290fxl9nt9eXdudCPYdac5FbvLbCzGwqdB3k4mflt6HmQn1/hsJ7b/b41Fzk4jUsCjUXyvvT6w8enif/xfuT/JdNmNndu3dtc3Mz/B8HADyetra2bGNj42P/96FQyLLMdnZ2rN1uW5LEy+gAAI+HPM+t0+nY+vq6FX7BXVooFAAAvx34fzQDAByhAABwhAIAwBEKAABHKAAAHKEAAHCEAgDA/Q8r58uRxbntrgAAAABJRU5ErkJggg==\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "image_id = 541\n",
        "pixel = np.array([17, 18, 185, 36, 215])\n",
        "model = resnet\n",
        "\n",
        "true_class = y_test[image_id, 0]\n",
        "print(class_names[true_class])\n",
        "prior_confidence = model.predict_one(x_test[image_id])[true_class]\n",
        "success = attack_success(pixel, x_test[image_id], true_class, model, verbose=True)\n",
        "\n",
        "print('Prior confidence', prior_confidence)\n",
        "print('Attack success:', success == True)\n",
        "helper.plot_image(perturb_image(pixel, x_test[image_id])[0])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FNWb8tOo-3zO"
      },
      "source": [
        "### Attack Function"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_FGxdjs1-3zO"
      },
      "source": [
        "Finally, we arrive at the attack itself: how do we find the pixels that will result in a successful attack? First, formulate it as an optimization problem: in an untargeted attack, minimize the confidence of the correct class, and in a targeted attack, maximize the confidence of a target class. This is precisely our `predict_class` function.\n",
        "\n",
        "When performing black-box optimizations such as the one pixel attack, it can be very difficult to find an efficient gradient-based optimization that will work for the problem. It would be nice to use an optimization algorithm that can find good solutions without relying on the smoothness of the function. In our case, we have discrete integer positions ranging from 0 to 31 and color intensities from 0 to 255, so the function is expected to be jagged.\n",
        "\n",
        "For that, we use an algorithm called [differential evolution](https://en.wikipedia.org/wiki/Differential_evolution). Here's an example of differential evolution optimizing the [Ackley function](https://en.wikipedia.org/wiki/Ackley_function) (if you're using Google Colab, run the code cell below):\n",
        "\n",
        "<br>\n",
        "\n",
        "![Ackley GIF](images/Ackley.gif)\n",
        "\n",
        "<br>\n",
        "\n",
        "Differential evolution is a type of evolutionary algorithm where a population of candidate solutions generate offspring which compete with the rest of the population each generation according to their fitness. Each candidate solution is represented by a vector of real numbers which are the inputs to the function we would like to minimize. The lower the output of this function, the better the fitness. The algorithm works by initializing a (usually random) population of vectors, generating new offspring vectors by combining (mutating) individuals in the population, and replacing worse-performing individuals with better candidates.\n",
        "\n",
        "In the context of the one pixel attack, our input will be a flat vector of pixel values:\n",
        "\n",
        "$$X = (x_1, y_1, r_1, g_1, b_1, x_2, y_2, r_2, g_2, b_2, ...)$$\n",
        "\n",
        "These will be encoded as floating-point values, but will be floored back into integers to calculate image perturbations. First we generate a random population of $n$ perturbations\n",
        "\n",
        "$$\\textbf{P} = (X_1, X_2, \\dots, X_n)$$\n",
        "\n",
        "Then, on each iteration we calculate $n$ new mutant children using the formula\n",
        "\n",
        "$$X_i = X_{r1} + F (X_{r2} - X_{r3})$$\n",
        "\n",
        "such that\n",
        "\n",
        "$$r1 \\neq r2 \\neq r3$$\n",
        "\n",
        "where $r1,r2,r3$ are random indices into our population $\\textbf{P}$, and $F = 0.5$ is a mutation parameter. Basically, we pick 3 random individuals from the previous generation and recombine them to make a new candidate solution. If this candidate $X_i$ gives a lower minimum at position $i$ (i.e., the attack is closer to success), replace the old $X_i$ with this new one. This process repeats for several iterations until our stopping criterion, `attack_success`, which is when we find an image that successfully completes the attack.\n",
        "\n",
        "<br>\n",
        "\n",
        "See [this excellent tutorial post](https://pablormier.github.io/2017/09/05/a-tutorial-on-differential-evolution-with-python/) on how differential evolution works in greater detail.\n",
        "\n",
        "We will use a [slight modification](differential_evolution.py) of [Scipy's implementation of differential evolution](https://docs.scipy.org/doc/scipy-0.17.0/reference/generated/scipy.optimize.differential_evolution.html) to utilize GPU parallelism by batching predictions together."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "tlO00HlxbB0k"
      },
      "outputs": [],
      "source": [
        "# Run this cell if you are using Google Colab to see the Ackley GIF\n",
        "if in_colab:\n",
        "    from IPython.display import Image\n",
        "    with open('images/Ackley.gif','rb') as file:\n",
        "        display(Image(file.read()))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 34,
      "metadata": {
        "id": "7rlzqXpqSgrW"
      },
      "outputs": [],
      "source": [
        "def attack(img_id, model, target=None, pixel_count=1,\n",
        "           maxiter=75, popsize=400, verbose=False):\n",
        "    # Change the target class based on whether this is a targeted attack or not\n",
        "    targeted_attack = target is not None\n",
        "    target_class = target if targeted_attack else y_test[img_id, 0]\n",
        "\n",
        "    # Define bounds for a flat vector of x,y,r,g,b values\n",
        "    # For more pixels, repeat this layout\n",
        "    bounds = [(0,32), (0,32), (0,256), (0,256), (0,256)] * pixel_count\n",
        "\n",
        "    # Population multiplier, in terms of the size of the perturbation vector x\n",
        "    popmul = max(1, popsize // len(bounds))\n",
        "\n",
        "    # Format the predict/callback functions for the differential evolution algorithm\n",
        "    def predict_fn(xs):\n",
        "        return predict_classes(xs, x_test[img_id], target_class,\n",
        "                               model, target is None)\n",
        "\n",
        "    def callback_fn(x, convergence):\n",
        "        return attack_success(x, x_test[img_id], target_class,\n",
        "                              model, targeted_attack, verbose)\n",
        "\n",
        "    # Call Scipy's Implementation of Differential Evolution\n",
        "    attack_result = differential_evolution(\n",
        "        predict_fn, bounds, maxiter=maxiter, popsize=popmul,\n",
        "        recombination=1, atol=-1, callback=callback_fn, polish=False)\n",
        "\n",
        "    # Calculate some useful statistics to return from this function\n",
        "    attack_image = perturb_image(attack_result.x, x_test[img_id])[0]\n",
        "    prior_probs = model.predict_one(x_test[img_id])\n",
        "    predicted_probs = model.predict_one(attack_image)\n",
        "    predicted_class = np.argmax(predicted_probs)\n",
        "    actual_class = y_test[img_id, 0]\n",
        "    success = predicted_class != actual_class\n",
        "    cdiff = prior_probs[actual_class] - predicted_probs[actual_class]\n",
        "\n",
        "    # Show the best attempt at a solution (successful or not)\n",
        "    helper.plot_image(attack_image, actual_class, class_names, predicted_class)\n",
        "\n",
        "    return [model.name, pixel_count, img_id, actual_class, predicted_class, success, cdiff, prior_probs, predicted_probs, attack_result.x]"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from differential_evolution import differential_evolution"
      ],
      "metadata": {
        "id": "bgeYRYXBHr8a"
      },
      "execution_count": 36,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pm_GwVzq-3zO"
      },
      "source": [
        "#### Untargeted Attack"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Y9OpBfzY-3zO"
      },
      "source": [
        "Let's look at one iteration of the untargeted attack. Here we will demonstrate a successful attack an image of a frog with the `resnet` model. We should see the confidence in the true class drop after several iterations.\n",
        "\n",
        "Try to see if you can successfully attack other images/models. The more pixels we are allowed to modify, the more likely it is we are to find a solution for any given image."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 37,
      "metadata": {
        "scrolled": false,
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "8C9-t4td-3zO",
        "outputId": "6fd53020-e708-41ac-f82b-db41ffed15a9"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "4/4 [==============================] - 2s 393ms/step\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.99873334\n",
            "4/4 [==============================] - 2s 396ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.88985825\n",
            "4/4 [==============================] - 2s 395ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.88985825\n",
            "4/4 [==============================] - 3s 671ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.6378472\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.6378472\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.60630155\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.5703629\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.5703629\n",
            "4/4 [==============================] - 3s 682ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.5703629\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.5703629\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.5703629\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "Confidence: 0.5703629\n",
            "4/4 [==============================] - 2s 483ms/step\n",
            "1/1 [==============================] - 0s 48ms/step\n",
            "Predicted Class: dog\n",
            "Confidence: 0.40759176\n",
            "1/1 [==============================] - 0s 68ms/step\n",
            "1/1 [==============================] - 0s 45ms/step\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "image_id = 102\n",
        "pixels = 1 # Number of pixels to attack\n",
        "model = resnet\n",
        "\n",
        "_ = attack(image_id, model, pixel_count=pixels, verbose=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "azvqD0iS-3zP"
      },
      "source": [
        "#### Targeted Attack"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jVWooBRa-3zP"
      },
      "source": [
        "In the targeted attack, we can choose which class we want a model to classify an image as. The task is much harder for the targeted attack, as we constrain the misclassification to a given class rather than any class that's not the correct one. We should see the confidence in the target class rise after several iterations.\n",
        "\n",
        "Below we try to cause the `lenet` to classify an image of a `ship` as an `automobile`. Try to change the parameters and see what happens."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0NciMf84-3zP"
      },
      "source": [
        "### Collect Results"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mHhrnGaR-3zP"
      },
      "source": [
        "Armed with all the necessary tools to conduct a one pixel attack, the final step is to collect relevant statistics on the targeted and untargeted attack. The relevant data points are what percentage of images were we able to successfully attack for a given model, and how the number of pixels affect this percentage.\n",
        "\n",
        "We will loop through every combination of all models, perturbations of 1,3,5 pixels, images, and target classes (for the targeted attack). This will take a lot of computational resources and time, so [skip to the statistics section](#Attack-Statistics) if that's not your idea of fun."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 40,
      "metadata": {
        "id": "uDH6vqTS-3zP"
      },
      "outputs": [],
      "source": [
        "def attack_all(models, samples=500, pixels=(1,3,5), targeted=False,\n",
        "               maxiter=75, popsize=400, verbose=False):\n",
        "    results = []\n",
        "    for model in models:\n",
        "        model_results = []\n",
        "        valid_imgs = correct_imgs[correct_imgs.name == model.name].img\n",
        "        img_samples = np.random.choice(valid_imgs, samples, replace=False)\n",
        "\n",
        "        for pixel_count in pixels:\n",
        "            for i, img_id in enumerate(img_samples):\n",
        "                print('\\n', model.name, '- image', img_id, '-', i+1, '/', len(img_samples))\n",
        "                targets = [None] if not targeted else range(10)\n",
        "\n",
        "                for target in targets:\n",
        "                    if targeted:\n",
        "                        print('Attacking with target', class_names[target])\n",
        "                        if target == y_test[img_id, 0]:\n",
        "                            continue\n",
        "                    result = attack(img_id, model, target, pixel_count,\n",
        "                                    maxiter=maxiter, popsize=popsize,\n",
        "                                    verbose=verbose)\n",
        "                    model_results.append(result)\n",
        "\n",
        "        results += model_results\n",
        "        helper.checkpoint(results, targeted)\n",
        "    return results"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 41,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "_6ZBl-wa-3zP",
        "outputId": "cd483d1e-cdaa-410f-b99f-cbf78871c88d"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            " resnet - image 8497 - 1 / 100\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: horse\n",
            "1/1 [==============================] - 0s 27ms/step\n",
            "1/1 [==============================] - 0s 40ms/step\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            " resnet - image 5313 - 2 / 100\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "4/4 [==============================] - 2s 621ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 678ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 535ms/step\n",
            "1/1 [==============================] - 0s 46ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 688ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 516ms/step\n",
            "1/1 [==============================] - 0s 43ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 655ms/step\n",
            "1/1 [==============================] - 0s 140ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 548ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 683ms/step\n",
            "1/1 [==============================] - 0s 48ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 489ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 392ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 395ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 430ms/step\n",
            "1/1 [==============================] - 0s 41ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 646ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 398ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 667ms/step\n",
            "1/1 [==============================] - 0s 42ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 643ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 685ms/step\n",
            "1/1 [==============================] - 0s 50ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 605ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 27ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 591ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 712ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 38ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 393ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 676ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 656ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 691ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 390ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 403ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 449ms/step\n",
            "1/1 [==============================] - 0s 45ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 683ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 672ms/step\n",
            "1/1 [==============================] - 0s 27ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 395ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 686ms/step\n",
            "1/1 [==============================] - 0s 48ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 27ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 685ms/step\n",
            "1/1 [==============================] - 0s 50ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 473ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 418ms/step\n",
            "1/1 [==============================] - 0s 57ms/step\n",
            "Predicted Class: cat\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "1/1 [==============================] - 0s 52ms/step\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            " resnet - image 1894 - 3 / 100\n",
            "4/4 [==============================] - 3s 681ms/step\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 669ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 397ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 488ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 679ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 392ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 644ms/step\n",
            "1/1 [==============================] - 0s 56ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 682ms/step\n",
            "1/1 [==============================] - 0s 36ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 45ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 666ms/step\n",
            "1/1 [==============================] - 0s 42ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 434ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 390ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 452ms/step\n",
            "1/1 [==============================] - 0s 53ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 693ms/step\n",
            "1/1 [==============================] - 0s 43ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 378ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 392ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 497ms/step\n",
            "1/1 [==============================] - 0s 42ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 689ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 379ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 501ms/step\n",
            "1/1 [==============================] - 0s 50ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 682ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 46ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 390ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 693ms/step\n",
            "1/1 [==============================] - 0s 45ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 400ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 681ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 36ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 667ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 45ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 686ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 660ms/step\n",
            "1/1 [==============================] - 0s 45ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 672ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 487ms/step\n",
            "1/1 [==============================] - 0s 46ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 684ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 3s 670ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 394ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: cat\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 27ms/step\n",
            "Predicted Class: cat\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            " resnet - image 4810 - 4 / 100\n",
            "4/4 [==============================] - 3s 682ms/step\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 495ms/step\n",
            "1/1 [==============================] - 0s 48ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 695ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 401ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 36ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 27ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 526ms/step\n",
            "1/1 [==============================] - 0s 60ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 634ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 48ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 679ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 395ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 669ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 479ms/step\n",
            "1/1 [==============================] - 0s 43ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 682ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 410ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 651ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 667ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 489ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 688ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 402ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 634ms/step\n",
            "1/1 [==============================] - 0s 51ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 679ms/step\n",
            "1/1 [==============================] - 0s 38ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 667ms/step\n",
            "1/1 [==============================] - 0s 51ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 630ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 686ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 480ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 614ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 685ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 685ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 27ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 502ms/step\n",
            "1/1 [==============================] - 0s 51ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 720ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 656ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 522ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 3s 699ms/step\n",
            "1/1 [==============================] - 0s 46ms/step\n",
            "Predicted Class: dog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: dog\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAGpCAYAAABiTSoFAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAn0ElEQVR4nO3de5DddX3/8df3XPfs7tlbdrObzZVkc7XIT+BXFfCHCI62HX9g51cZWgfSWmmrjpGLor2og622FfzRlinQqRCg0qkjl6lUcYAKhTCGi0QLJCEBQm6b++5mb+f6/fz+yC+fn0sMvt/+goB9PmacMdl33vvNOd9zXudk+b5OEkIIAgBAUub1PgAAwBsHoQAAiAgFAEBEKAAAIkIBABARCgCAiFAAAESEAgAgylmG0jTV7t27VS6XlSTJa31MAIATLISg8fFxDQ4OKpM5/vsBUyjs3r1b8+fPP2EHBwB4fezYsUPz5s077tdNoVAulyVJ511wpfL5oukbb395l2lOkvpmd5tnJal3Vod5tjZdce0uFG1/P0nKOmYlad/eA+bZyeq4a3dLm+9YKpMN82ytWnXtfpUXIT9leda1W037aEh8xx2Cbz7JpObZTOL7e6Z1x2753r2Hhr3ZJk3rrt1JYr+D6g37OShJhULedyyO2an6QdfujOznSk6zXLuV6TGPlrrbzLONRk1PPnJ7fD4/HlMoHP0no3y+qHy+xXQA2VzBNCdJOWPQHJUv2I5BkkLTV+3k2Z0r+I7b8/fMpb4nKO9tmMvbn7mbztvQFQqpMxQcj/SQ+I47BPsTsfTahkLiOBZ3KDjqzrz/Wpwk9js/Db7l2awzFBzrs6npqTDKOF6dZOU7bmUcz52O59mjftaPAPhBMwAgIhQAABGhAACICAUAQEQoAAAiQgEAEBEKAICIUAAARK4rNsZGR80XSb1at8YrVSu+C7X27N1nni23d7p2J46rPfft3+3b7Zjt6/JdBVmtTbrmm037fD6ddu0uBPuFWi0534U9QTX77pJrtfJF38Vrmaz9Hp2q+S6OGp20X0k8UfUd97TjArNMxnfRXSa1H0sanBdeJb7bsJ7az3HfZY5ScPyJIN9V4RnHRZGJ4zZJEtte3ikAACJCAQAQEQoAgIhQAABEhAIAICIUAAARoQAAiAgFAEBEKAAAIkIBABC5rhuvVCrKNmyXSpfb2817PZd1S5LjKn01gu/DwTOO6917Zr36B2Afo2n/XNc2VVyrc+l+13zvgP1G7HJ8brUklRyvNZKmvbZCksrtrebZWb3drt31hu82l+zn7fSk7++5Y8Q++8Ko/bySpM377LUl41VfAUTqeEoptNg/dF6SlPU9lhvNMfuw88OoU8fnloeMr+Yil7PfhvWa/b5vNGyzvFMAAESEAgAgIhQAABGhAACICAUAQEQoAAAiQgEAEBEKAICIUAAARIQCACAiFAAAkav7qBkkGSs/ao1J896+Ll9Hzfx588yzLS1F1+6QVM2zzYkDrt2lyXHzbH9m1LX7pMW+27Cl3GUfrvv6bzpb7P1Eu7Zvc+0u5grm2Zam7zXPnmFHV46k1vaSeXblYI9r98q59k6b4UlfZ9O2sU7z7NMv289ZSXp215R5Ns357p9a1dfxFBznbSb4uo88T50h+LrdPM+dIbE/HpoNW3cU7xQAABGhAACICAUAQEQoAAAiQgEAEBEKAICIUAAARIQCACAiFAAAEaEAAIh8NReNVJLtku2uDnsFwML59toKSWptaTPP7t+/y7e7MG2efcuA/e8oSXO77Td3a9VeFSFJ9aavzuMH618wz+4+OOHanXfUF0yOOXdn7X/PfKHFtXt4dNQ139FVNs8e2G+vT5Gk5fPt9//sPt952N5mr0ZYMH+Ba/fszXvMsw9vsJ+DkjRZ9dVFZD2veZ01F0mStc+6NkvNUDfPFgt582wmY6v94J0CACAiFAAAEaEAAIgIBQBARCgAACJCAQAQEQoAgIhQAABEhAIAICIUAAARoQAAiFzdR8VCQbmcrXtmoHfAvPfwIV//zcZdW8yzBUeXkSS9Y8Vs8+xA1tdn053aj+XAeMW1++kX97nmNx2w3+Z7J6Zcu5NM0zzbTO09L0d222/DTN51equZ8XXrTE6Pm2efesF3f1am282zq5q9rt25ov027O6394xJ0qoBe2fT3n5fZ9OPXtrrmq87erKU8b0+Dk1bj5Akpan98SBJmcTelhSC/Tiss7xTAABEhAIAICIUAAARoQAAiAgFAEBEKAAAIkIBABARCgCAiFAAAESEAgAgcvUA9HR0K5+3XTo+enDUvHfkwGHPYSiTzZpnF86x11ZIUtlx+Xp1/4hr93TBvnvLnoOu3Rt2DLvmJ/Jl82ytxVcXEZqO6gr7Ff2SpIYa5llvhUbe+RIpm9j/wObRSdfu8clD5tmxMd+5snLxXPNsqbPPtbtcajHPLuzvcu0eHrXXikjSjhFHJYrjvpQk2R/K7nM8Te11K57nwhBss7xTAABEhAIAICIUAAARoQAAiAgFAEBEKAAAIkIBABARCgCAiFAAAESEAgAgIhQAAJGr1CZNUzWNvRyjI2PmvZnE161TLrWZZ1tSX/FI5fCUebbW4tt9uOmYrdg7fiSpGmqu+amGvUemlvHdP0lqL4bJBHtXjiSFjON1TOq7TdTw3eaNhr1bqel8+ZVpazfPltp6XLubqf3+HJvwdTbVS63m2fY2+6wkLRqY5ZrfN7bdPFs39gJFGUfnkOxdRpLkqWFq77Dfho063UcAACdCAQAQEQoAgIhQAABEhAIAICIUAAARoQAAiAgFAEBEKAAAIkIBABC5+gsOj40plyuaZquOCoBs4uh/kJTL2o5BklTxVR1UmhXzbLbTXrchSbmM/TaZ1z/g2v38vsOu+YOT9vlG8NV55FLH/ZMpuHYrsVdoSL7qgnrWdx6mwT5fzPqqQqZr9sqNjPM2LDsqNBr1adfuqbq90qExbX88SNKS2b6ai00v2msu9k75Kk6yjodEaqwGOqqlYD9v+2bbK07qNdtzG+8UAAARoQAAiAgFAEBEKAAAIkIBABARCgCAiFAAAESEAgAgIhQAABGhAACICAUAQOQqZGlUKwpNW4/H4OBs897urpLnMNSa2PuJktGdrt39/fYukf7eDtfurKNHZs+Yr3NmdHrKNV/P2O/6RL7uo9bU3gm0qKvLtbvcab/Ndx3a69q9e+KAa77mqFZy1kdp/7T9/n9xeI9r91uX9Zln83nfgderefPs2L7drt0r3jLkml80aP977tls70mSpCRnf/xkEl/3URLs/V71StU+W7PN8k4BABARCgCAiFAAAESEAgAgIhQAABGhAACICAUAQEQoAAAiQgEAEBEKAICIUAAARK7uo0wxp4yx8yOXsfd3DPa0eA5DsxzjnX32DiZJGhroNs925Hy9MNVx+3zijOvg/APZpGCeLTR8x/KOk5aYZ//nqWe5dne0tJtntx3wdQLd/+PHXfMvje83z04l9seDJE1W7fNpzn5fSlK2YD8PuzvtXUaSVE7s988LL9VduxuTI675BY4esx9u2ebanch+7N7uME+f0Z5d9n6vRqNmmuOdAgAgIhQAABGhAACICAUAQEQoAAAiQgEAEBEKAICIUAAARIQCACAiFAAAkavmor27Tbl80TQ7OnLAvHd3wz4rSUtPmWeend3hq9DoKNlzMtfwXaafayuZZ3u6u1y724r23ZJ0aKpp353x1SisGJxvnl06d9C1uyTb+SdJOV+zhB5zVH9IUnupbJ6tT064dudT+2xPd6drd7mjwzw7MNDq2j1RsXeivPP05a7dBw+NuuZnd7aZZ7tafff92HTFMZ117W427DUXlalJx15qLgAAToQCACAiFAAAEaEAAIgIBQBARCgAACJCAQAQEQoAgIhQAABEhAIAICIUAACRq/to+cJ+FYq2LqFqpdd+EOOHPIeh9mZini1n7LOSlNSmzbO1qqf/RJoz194J1DEw17V7POfrbrnrofXm2T07Drp2P/n8ZvPsiiUrXLsX9Mwxz/5o+zbX7h/t2uGa356xnyulxNY7c9T/eNtS8+z/Ou9M1+55ZXtPVpr6jjubjJln+zt9nUCdpR7XfGnU3sO0am6/a/f6TfZzpZFzPgdl7M8racV+e6dNW1cb7xQAABGhAACICAUAQEQoAAAiQgEAEBEKAICIUAAARIQCACAiFAAAEaEAAIhcNRcTY1MqFFLTbKFkX93a6sumlpamebaYsR3vUZOjI+bZ9lLetbuYtV1mLkn9/a2u3Rcu8lUdDK1cbJ793oOPu3Zvee5l8+z//ubtrt2ZFnudx4FxXz1HM+87V85Ytsg8+54zf8W1+13/bZl5tt/5+Kkdtp/jh8fsVR6S1Npir3SohODaXW/47p9iYn+eWDTgq7nYuuuweXbvhH1WknI5+3NnW0vZPNto2CpLeKcAAIgIBQBARCgAACJCAQAQEQoAgIhQAABEhAIAICIUAAARoQAAiAgFAEBEKAAAIlf30aGREeXzRdPsoiXzzHvndPZ6DkP5woR5dnJy3LW7mNi7W8odba7d0xV7B0oyZj8OSWrLVF3z71gx2zx78oqLXLuf3Tpsnn3qh8+4dtdl77PZe2CPa/fbVq5yzZ9zxtvNs51d9uOWpGKYNM+mzo6nicn95tlGw3deWZ8fJCmX8T1+Do8ccs2nqb1bqS3n62F6q6P36qUDvvtnx357N1Vbq70jrdGwPd3zTgEAEBEKAICIUAAARIQCACAiFAAAEaEAAIgIBQBARCgAACJCAQAQEQoAgMhVc9HRUVK+YLuMfWTEfmn3wg5fzcXh8WnzbL5Rc+1eONhlnu3sneXa3VTDPNvIpK7d09NjrvkksdcutJZ9FQAnrxowz656q70ORZJyGfvrmLER323S2z/HNZ8p2h8+lcO+yo10yn6OTx0ade2uTtmrK1rb7DUKkpS0lMyz0xXXak2m9poYSRqftFeFFFzPhFJrvm6e/e9vW+Ha3b37gHn2+a32c7zRsB0z7xQAABGhAACICAUAQEQoAAAiQgEAEBEKAICIUAAARIQCACAiFAAAEaEAAIgIBQBA5Gr8mL9gvopFW7fJ/gP7zHsnRu0dJZJ0oDZlnu3tSly7k2LBPBtyvkxtLbXZh7NZ1+5szjevgn0+zfp6mEptecdx2G9vSUocf81iW7dvt3w9WfVp+3mYqfluw4lD9k6bqcO+jqeio+gn43g8SFLD0U01PuXsMpryPU/sH58wz07X7V1GklSv2IubWjL2njFJOv3UlebZvYeeNc/W67YHD+8UAAARoQAAiAgFAEBEKAAAIkIBABARCgCAiFAAAESEAgAgIhQAABGhAACICAUAQOTqPtq9a48KhaJptq+v17x3dP9Oz2Eobdg7akotjh4eSZOVhnl2qlp17W7vbjHPZuTrbErl69ZJFcyznq4cScpk7Mder9k7ZCQpyTvKjxLna56Gr/8mcXTxZJy9PaljvujosZKkjs6yebYWfOdhddJ+Gx7aP+raPVXxPd6ybfbepuoB3/3T6umPqvvO8WzDPj+r096nVqvZngt5pwAAiAgFAEBEKAAAIkIBABARCgCAiFAAAESEAgAgIhQAABGhAACICAUAQOTqLxg5OKp83lZzMTVuv1R7sKvkOQzVUnvNRaXpu0w/W7RXAEzW7ZUYkhTy9svuQ7DXUEhSkvrmM47L9BvOCo16Zco820x9u1uyHebZYrHdtTukvjqCWjhs3x18FRrFkv1cSRu+x0+at1e/JE3f68ZazV4XMTXZdO0eG/VVURTb7LUyLXnnedhmr5doLfqegyYO7TPPpuNj9tm67XmTdwoAgIhQAABEhAIAICIUAAARoQAAiAgFAEBEKAAAIkIBABARCgCAiFAAAESEAgAgcnUftbeVzd1HE2P2/puWwUHPYWj8wIh5Ntdq7z+RpHJPt3k2m0y7do87qls6OuwdTJKULfryPWmx3Y+SlGm33yaSlCvY+4m88nnP7eLrBGo4uowkqZa1984E53mYL/baj+Ow/TgkqdHImmcrTV+/V7NgP69C4rtN6lO+/qiFg/3m2da59ttbkpqOTqhyp++xfLhm74TastV+ztaytvuddwoAgIhQAABEhAIAICIUAAARoQAAiAgFAEBEKAAAIkIBABARCgCAiFAAAESumotMJqdsJm+arUxWzHtf2LLFcxjq77FfBr7wpEWu3a1t9sv0u8qdrt2ewoBMW49rd65gu1+OSvP2qoNs0XcsmaL9dglJ4tot46X6ktQM9llJSgqzXPPt7fY6j7Q54dpdmLZXV1Qyu1y7x/buM89ON2uu3Z76lLHDvtukNuk7ltZMq3m2LVtw7a4n9kdzLgmu3QsGZ5tn3/UO+3FPVSq66c6fPcc7BQBARCgAACJCAQAQEQoAgIhQAABEhAIAICIUAAARoQAAiAgFAEBEKAAAIkIBABC5uo9CCEqDrcej3rD3lLS12DtKJGnF0BzzbKHg69ZpJvZepaTo6xvq7nF0CCWuu0a11NOsJOUcXS+ZfItrdyZr779R1ncbhozj/gmpb3fiO5ZspmyezSf2nqQjx2K//9OCvSdJklScNI92l3y9V3v2jZpn9+3zHXet6ntM7Hz5kHm2s+zrPsoU7OdWGPU9B3XV7bMFRxdYI2N77uadAgAgIhQAABGhAACICAUAQEQoAAAiQgEAEBEKAICIUAAARIQCACAiFAAAkeu68enpKTUatpqBNLFfq93R5qsXaE3sl5h3OCs0Wtrs8xP1imt3oWmvoii2OKoiJKUNX6VDXbZL3iUpn7NfSi9JIW8/rZLE9/dMHDUkGfmqP9KG/TaRJAX7sYSsr+qg2rS/XmsU2l27e+ba6zkO7tnr2r1t2wHz7KFx3znbUux2ze/fa6/zGN034trdOavNPFvLOHorJI0etp9X9Yz9OKarVdMc7xQAABGhAACICAUAQEQoAAAiQgEAEBEKAICIUAAARIQCACAiFAAAEaEAAIgIBQBA5Oo+mpweV65u68/IFe15kwm+DqH6yEHzbKnR79pdlL3nZ3RywrW72mLvYmkpdrp2lzq6XPOpo8+oHgqu3TnZ+4zyWWfHU3D0yKS+bp2MfPOe/cHReyVJIbX335Q6Ony7x6bNs7t3+rqP9u61dx/lWn2dTYf3+x5vLcauH0kKU/bHpiRVp+3nYS3ru+8zLfbjPlw/ZJ6t1mu272/eCAD4pUcoAAAiQgEAEBEKAICIUAAARIQCACAiFAAAEaEAAIgIBQBARCgAACJfzcXkpHI52+XdfV0l8962Fs9RSGrYL0mvjNgvA5ekWq/9YCoVe12AJG1/Ybt5tmPEV/3RveAk13zvwqXm2TTrvIOSvH13Gny7g6NaItirPCRJie9YEs9s0zMtlQr2x8/UyLhr9/5tL9tnd+5y7Z467KiiyPhu75GJMdd8a81eFZKv22clqT5hr6KoZ3y7m1P2c3zPmOM4mrbnbt4pAAAiQgEAEBEKAICIUAAARIQCACAiFAAAEaEAAIgIBQBARCgAACJCAQAQEQoAgMjVfTQ40K18vmia7cnb+z76u3zdOqVQM8/WqlOu3W1trY5h221x1Eub7N1HaeLreQntvvnygP31QKGt4NqdJPb73tcI5BPk6z4Kie9oQmK/DZOM7/VX5bC9V2ty/37X7vq4vQ9s7MAB1+7QtN/mO4eHXbt3jR52zRca9vuzkPM9ljOO3qZ6Yu8ykqSmYz5tbbcfR4PuIwCAE6EAAIgIBQBARCgAACJCAQAQEQoAgIhQAABEhAIAICIUAAARoQAAiAgFAEDk6j7q7y2qULB1hGQm7T0lLa6jkIqOw67WG67dpRZ795GrJ0nSnlZ7j0y+1OPaHdThmh/Zb+9Kam/ae14kKe+osioWfZ0zieN1TLNu72CSpOAbV9bxkmpqyt43JEkHd201z87K+bp19k7Y7/v9zr6h0bTTPLtnzN5hJknVYrdrvtlqPxEbuTbX7npq6xGSpImK775P8vbnt75Zs82zad12e/NOAQAQEQoAgIhQAABEhAIAICIUAAARoQAAiAgFAEBEKAAAIkIBABARCgCAyFUw8dZTTlepZKt2GH7+WfPeNPgupff0C+w/OOJaPXZw3Dw72NXr2j0wZ555dqTm6IqQ1Ex8dRH7du0xz+7YucO1O9dqP5b+gUHX7nJbu3l2eMfLrt0To75zpavDXo2QTe3VEpLUHqrm2cP7fMe9/vHnzLM7D/m6P/Y3psyz5R7740GSZpV9NReZQt4826j6qly2v/iSeba93Xfc7eWyfbiZNY9mjHU1vFMAAESEAgAgIhQAABGhAACICAUAQEQoAAAiQgEAEBEKAICIUAAARIQCACAiFAAAkav7aKpSUEgKptmxaXvedHQ6uj4kDb1lpXn2pWd/7Nr93IYfmmdbS/ZuFUmqTEybZ0er9u4bSapNpa75fGK/f/LFkmu3cvZOoE3P7XStrlcq5tmtmza4dpeLvv6bVUNzzbNzfKe4Jg7vN8/+4LGnXLt3H7Dfhmrvd+1W1X4eDgz4usPqqa+HaXzK3ql2aHzUtbtvYZd5dvZs39/zxa3bzbMjh+xdU81G3TTHOwUAQEQoAAAiQgEAEBEKAICIUAAARIQCACAiFAAAEaEAAIgIBQBARCgAACJXzcVjjz6mfL5omj00fsi8d0e76zDUPWvQPDv/pFWu3dmJPebZ7Rs3uXY32+yXuw8f8tVWvHzAVxfRPctx6X0m69qdK9grGvbtGXPt3vaivQKgXPK95nnHKUOu+fZMq3l25+ZnXbuLtoeZJGnV6e9w7Z79K/bakr3Tvvv+pb32+/7l7b5zdmzSXukgSWPjk+bZ2mTNtfuUlcvNs7mS7zbc1Wq/83NTjmqWhu04eKcAAIgIBQBARCgAACJCAQAQEQoAgIhQAABEhAIAICIUAAARoQAAiAgFAEBEKAAAIlfpUK0pBWOM5PP2Xpi9B0Y9h6F7vrfePDtvdodr9+J+ey9M76y8a3dR9p6SidS3Wzn77S1Ju/eOmmf3HbT3WEnS3n0HzLOVSsO1u9hqv39Gq45eGElTTz3tmn/qx/bZ1qTu2j1v7oB5tr3H15P1xLP2HqbhQ76+Ic+9OV2puHaXWttd8329nebZ8lxH2ZSknr5Z5tkXd7zs2h2y9qflYrv9NsnWq6Y53ikAACJCAQAQEQoAgIhQAABEhAIAICIUAAARoQAAiAgFAEBEKAAAIkIBABC5ai5CNlHIJrbhhr1ioLvPfkm/JBWyWfNs6PTVPxTnzzHPjqXTrt1ThyfNsx09/a7ds9rbXPMjY+Pm2ULNfntLUkvFXtHR0WOvC5CkgqPmouK8f0KxxTW/23EbpnVfFcWLm/eZZycnfTUKNcdrwVzR9/ip1ex1HvmCb3c256yVKdjnV7x1qWt3S8lei1Hbvt21uy7H461gfD6WpKRpGuOdAgAgIhQAABGhAACICAUAQEQoAAAiQgEAEBEKAICIUAAARIQCACAiFAAAkanmIoQjlRX1etW8uF6v2WczvmzKpPbLwKs13+7paXs1Qhp8NQqVuu0yc0nKT0+5dqfBV0VRqdiPvVatuHbXa/bzRKljVpJy9sv6a87dvhIFqeb4ewbHY0eSUtlrYjyPNUmqO14LhozvvKrX7TUXCvbHgyTlnC9hPe0s087HW5D92Gs13+On4ThXGnX746Hxf8+To8/nx5OEnzUhaefOnZo/f775mwMA3ph27NihefPmHffrplBI01S7d+9WuVxWkjgKmAAAbwghBI2Pj2twcFCZV/nXGVMoAAD+a+AHzQCAiFAAAESEAgAgIhSAn8NDDz2kJEk0Ojr6eh8KcEIRCnhDSJLkVf/3xS9+8fU+ROC/BNdnNAOvleHh4fj//+Vf/kWf//zntXnz5vh77e3t8f+HENRsNpXLcfoCJxrvFPCGMDAwEP/X2dmpJEnirzdt2qRyuazvfve7Ou2001QsFvXoo49q9erVuuCCC2bs+dSnPqV3v/vd8ddpmuorX/mKTjrpJJVKJZ1yyin61re+5T6+73znO1q2bJlKpZLOOeccbdu27ZiZO++8U295y1tULBa1aNEiXXvttTO+Pjw8rN/4jd9QqVTSSSedpDvuuEOLFi3Sdddd5z4e4LXCSy28aXz2s5/VNddco8WLF6u7u9v0Z77yla/on/7pn3TjjTdq6dKl+o//+A99+MMfVl9fn84++2xJ0qJFi7R69erj/hPVjh079Ju/+Zv6+Mc/rksvvVRPPvmkrrjiihkzTz31lD70oQ/pi1/8oi688EI99thj+tjHPqZZs2Zp9erVkqSLL75YBw4c0EMPPaR8Pq/LL79c+/bt+7lvD+C1QCjgTePqq6/We9/7XvN8tVrVl7/8ZT3wwAN65zvfKUlavHixHn30Ud10000xFJYsWaLe3t7j7rnhhhu0ZMmS+Mp/+fLl+s///E/91V/9VZz52te+pnPPPVd/9md/JklatmyZnnvuOX31q1/V6tWrtWnTJj3wwAN64okndPrpp0uS/vEf/1FLly713QjAa4xQwJvG0SdTq61bt2pqauqYIKnVanrb294Wf/3ggw++6p6NGzfq7W9/+4zfOxoyPzlz/vnnz/i9M888U9ddd52azaY2b96sXC6nU089NX59aGjI/I4H+EUhFPCm0dbWNuPXmUzmmMbHn2zpnJiYkCT927/9m+bOnTtjrlgsvkZHCby5EQp40+rr69Mzzzwz4/c2bNigfP5ICfaqVatULBa1ffv2+E9FP4+VK1fqX//1X2f83g9+8INjZtatWzfj99atW6dly5Ypm81q+fLlajQaevrpp3XaaadJOvJOZmRk5Oc+LuC1wH99hDet97znPXryySd12223acuWLfrCF74wIyTK5bKuvPJKXXbZZbr11lv1wgsv6Ic//KH+7u/+TrfeemucO/fcc3X99dcf9/v84R/+obZs2aJPf/rT2rx5s+644w6tXbt2xswVV1yhBx98UF/60pf0/PPP69Zbb9X111+vK6+8UpK0YsUKnXfeebr00kv1+OOP6+mnn9all16qUqlE8zDeWALwBnPLLbeEzs7O+Ovvf//7QVIYGRk5Zvbzn/986O/vD52dneGyyy4Ln/jEJ8LZZ58dv56mabjuuuvC8uXLQz6fD319feF973tfePjhh+PMwoULwxe+8IVXPaZvf/vbYWhoKBSLxfCud70r3Hzzzccc07e+9a2watWqkM/nw4IFC8JXv/rVGTt2794dfu3Xfi0Ui8WwcOHCcMcdd4TZs2eHG2+80XPzAK8pqrOB18nRD6964IEHdO65577ehwNI4vMUgF+Yf//3f9fExIROPvlkDQ8P6zOf+Yx27dql559/Pv4cBHi98YNm4BekXq/rj//4j/Xiiy+qXC7rjDPO0De+8Q0CAW8ovFMAAET810cAgIhQwJvKK0vw3v3ud+tTn/rUL/w4TtTnKaxdu1ZdXV0n5JiAE4FQwP+31atXx889KBQKGhoa0tVXX61Go/Gaf++77rpLX/rSl0yzfDAO8LPxg2acEO9///t1yy23qFqt6jvf+Y4+/vGPK5/P63Of+9wxs7VaTYVC4YR8356enhOyB8ARvFPACVEsFjUwMKCFCxfqj/7oj3TeeefFaoij/+TzF3/xFxocHNTy5cslHamk/tCHPqSuri719PTo/PPPn/E5Bc1mU5dffrm6uro0a9YsfeYznzmm6+iV/3xUrVZ11VVXaf78+SoWixoaGtLXv/51bdu2Teecc44kqbu7W0mSxEpry2cuWD5PwWLt2rVasGCBWltb9cEPflAHDx48ZuZoK2uhUNDy5ct1++23z/j6pk2bdNZZZ6mlpUWrVq3SAw88oCRJdM899/xcxwT8JEIBr4lSqaRarRZ//eCDD2rz5s26//77de+996per+t973ufyuWyHnnkEa1bt07t7e16//vfH//ctddeq7Vr1+rmm2/Wo48+qkOHDunuu+9+1e978cUX65//+Z/1t3/7t9q4caNuuukmtbe3a/78+brzzjslSZs3b9bw8LD+5m/+RtKRz1y47bbbdOONN+rZZ5/VZZddpg9/+MN6+OGHJf2/z1P4wAc+oA0bNuj3f//39dnPfvaY750kyTH1Fz9p/fr1+shHPqJPfOIT2rBhg8455xz9+Z//+YyZu+++W2vWrNEVV1yhZ555Rn/wB3+g3/3d39X3v/99SUeC8oILLlBra6vWr1+vf/iHf9Cf/Mmf/Ix7A3B4Ha+mxi+JSy65JJx//vkhhCO1Evfff38oFovhyiuvjF/v7+8P1Wo1/pnbb789LF++PKRpGn+vWq2GUqkUvve974UQQpgzZ07467/+6/j1er0e5s2bF79XCCGcffbZYc2aNSGEEDZv3hwkhfvvv/+nHudPq8uoVCqhtbU1PPbYYzNmP/KRj4SLLroohBDC5z73ubBq1aoZX7/qqquO2bV8+fJw1113Hfd2uuiii8Kv//qvz/i9Cy+8cEalxxlnnBE++tGPzpj5rd/6rfjnvvvd74ZcLheGh4fj1++///4gKdx9993H/d6AFT9TwAlx7733qr29XfV6XWma6rd/+7dnfJLZySefPOPnCD/60Y+0detWlcvlGXsqlYpeeOEFjY2NaXh4eMbnGORyOZ1++unH/BPSURs2bFA2m3U1olo+c8HyeQrSkX/WeTUbN27UBz/4wWP23HfffTNmLr300hkzZ555ZnxXs3nzZs2fP18DAwPx67/6q7/6qt8X8CAUcEKcc845uuGGG1QoFDQ4OKhcbuap9crPQpiYmNBpp52mb3zjG8fs6uvr+7mOoVQquf8Mn7kAzMTPFHBCtLW1aWhoSAsWLDgmEH6aU089VVu2bNHs2bM1NDQ043+dnZ3q7OzUnDlztH79+vhnGo2GnnrqqePuPPnkk5WmafxZwCsdfafSbDbj7/3kZy688jjmz58v6chnJTz++OMzdr3y8xQsVq5cOePv89P2HO9zGVatWiXpyEeB7tixQ3v37o1ff+KJJ9zHAhwPoYDXxe/8zu+ot7dX559/vh555BG99NJLeuihh/TJT35SO3fulCStWbNGf/mXf6l77rlHmzZt0sc+9rFXvcZg0aJFuuSSS/R7v/d7uueee+LOb37zm5KkhQsXKkkS3Xvvvdq/f78mJiZMn7lg+TwF6chnJrzaD8I/+clP6r777tM111yjLVu26Prrr5/xT0eS9OlPf1pr167VDTfcoC1btuhrX/ua7rrrrvi5DO9973u1ZMkSXXLJJfrxj3+sdevW6U//9E8lic9lwInxev9QA29+P/mDZs/Xh4eHw8UXXxx6e3tDsVgMixcvDh/96EfD2NhYCOHID5bXrFkTOjo6QldXV7j88svDxRdffNwfNIcQwvT0dLjsssvCnDlzQqFQCENDQ+Hmm2+OX7/66qvDwMBASJIkXHLJJSEE22cuWD5PQVK45ZZbXvW2+vrXvx7mzZsXSqVS+MAHPhCuueaaGT9oDiGEv//7vw+LFy8O+Xw+LFu2LNx2220zvr5x48Zw5plnhkKhEFasWBG+/e1vB0nhvvvue9XvDVhQiAe8ya1bt05nnXWWtm7dqiVLlrzeh4M3OUIBeJO5++671d7erqVLl2rr1q1as2aNuru79eijj77eh4ZfAvzXR8CbzPj4uK666ipt375dvb29Ou+883Tttde+3oeFXxK8UwAARPzXRwCAiFAAAESEAgAgIhQAABGhAACICAUAQEQoAAAiQgEAEBEKAIDo/wC2tP+SKMiVSQAAAABJRU5ErkJggg==\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            " resnet - image 654 - 5 / 100\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 37ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 52ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 680ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 379ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 668ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 602ms/step\n",
            "1/1 [==============================] - 0s 46ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 689ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 674ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 523ms/step\n",
            "1/1 [==============================] - 0s 53ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 597ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 36ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 466ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 679ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 27ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 687ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 399ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 41ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 700ms/step\n",
            "1/1 [==============================] - 0s 48ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 43ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 670ms/step\n",
            "1/1 [==============================] - 0s 48ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 394ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 43ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 692ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 455ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 697ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 42ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 43ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 665ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 40ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 27ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 672ms/step\n",
            "1/1 [==============================] - 0s 43ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 582ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 637ms/step\n",
            "1/1 [==============================] - 0s 51ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 625ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 685ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 399ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 390ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            " resnet - image 6975 - 6 / 100\n",
            "4/4 [==============================] - 3s 690ms/step\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: automobile\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            " resnet - image 9854 - 7 / 100\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 676ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 38ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 571ms/step\n",
            "1/1 [==============================] - 0s 55ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 3s 657ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: truck\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: truck\n",
            "2/4 [==============>...............] - ETA: 1s"
          ]
        },
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-41-53a559470347>\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0muntargeted\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mattack_all\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodels\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msamples\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m100\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtargeted\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;32m<ipython-input-40-67afb077e253>\u001b[0m in \u001b[0;36mattack_all\u001b[0;34m(models, samples, pixels, targeted, maxiter, popsize, verbose)\u001b[0m\n\u001b[1;32m     17\u001b[0m                         \u001b[0;32mif\u001b[0m \u001b[0mtarget\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0my_test\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mimg_id\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m                             \u001b[0;32mcontinue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 19\u001b[0;31m                     result = attack(img_id, model, target, pixel_count, \n\u001b[0m\u001b[1;32m     20\u001b[0m                                     \u001b[0mmaxiter\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mmaxiter\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpopsize\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mpopsize\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     21\u001b[0m                                     verbose=verbose)\n",
            "\u001b[0;32m<ipython-input-34-a3ba829b94fd>\u001b[0m in \u001b[0;36mattack\u001b[0;34m(img_id, model, target, pixel_count, maxiter, popsize, verbose)\u001b[0m\n\u001b[1;32m     22\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     23\u001b[0m     \u001b[0;31m# Call Scipy's Implementation of Differential Evolution\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 24\u001b[0;31m     attack_result = differential_evolution(\n\u001b[0m\u001b[1;32m     25\u001b[0m         \u001b[0mpredict_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbounds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmaxiter\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mmaxiter\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpopsize\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mpopmul\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     26\u001b[0m         recombination=1, atol=-1, callback=callback_fn, polish=False)\n",
            "\u001b[0;32m/content/drive/MyDrive/one-pixel-attack-keras-master/differential_evolution.py\u001b[0m in \u001b[0;36mdifferential_evolution\u001b[0;34m(func, bounds, args, strategy, maxiter, popsize, tol, mutation, recombination, seed, callback, disp, polish, init, atol)\u001b[0m\n\u001b[1;32m     20\u001b[0m                                          \u001b[0mcallback\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcallback\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     21\u001b[0m                                          disp=disp, init=init, atol=atol)\n\u001b[0;32m---> 22\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0msolver\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msolve\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     23\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     24\u001b[0m \u001b[0;32mclass\u001b[0m \u001b[0mDifferentialEvolutionSolver\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/content/drive/MyDrive/one-pixel-attack-keras-master/differential_evolution.py\u001b[0m in \u001b[0;36msolve\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    146\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mnit\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmaxiter\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    147\u001b[0m             \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 148\u001b[0;31m                 \u001b[0mnext\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    149\u001b[0m             \u001b[0;32mexcept\u001b[0m \u001b[0mStopIteration\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    150\u001b[0m                 \u001b[0mwarning_flag\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/content/drive/MyDrive/one-pixel-attack-keras-master/differential_evolution.py\u001b[0m in \u001b[0;36m__next__\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    219\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mtrial\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mtrials\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_ensure_constraint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrial\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    220\u001b[0m         \u001b[0mparameters\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_scale_parameters\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrial\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mtrial\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mtrials\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 221\u001b[0;31m         \u001b[0menergies\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mparameters\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    222\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_nfev\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0mitersize\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    223\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-34-a3ba829b94fd>\u001b[0m in \u001b[0;36mpredict_fn\u001b[0;34m(xs)\u001b[0m\n\u001b[1;32m     14\u001b[0m     \u001b[0;31m# Format the predict/callback functions for the differential evolution algorithm\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mpredict_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mxs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 16\u001b[0;31m         return predict_classes(xs, x_test[img_id], target_class, \n\u001b[0m\u001b[1;32m     17\u001b[0m                                model, target is None)\n\u001b[1;32m     18\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-20-713e560dbc26>\u001b[0m in \u001b[0;36mpredict_classes\u001b[0;34m(xs, img, target_class, model, minimize)\u001b[0m\n\u001b[1;32m      2\u001b[0m     \u001b[0;31m# Perturb the image with the given pixel(s) x and get the prediction of the model\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m     \u001b[0mimgs_perturbed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mperturb_image\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mxs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 4\u001b[0;31m     \u001b[0mpredictions\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mimgs_perturbed\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtarget_class\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      5\u001b[0m     \u001b[0;31m# This function should always be minimized, so return its complement if needed\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mpredictions\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mminimize\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0;36m1\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mpredictions\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/content/drive/MyDrive/one-pixel-attack-keras-master/resnet.py\u001b[0m in \u001b[0;36mpredict\u001b[0;34m(self, img)\u001b[0m\n\u001b[1;32m    175\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    176\u001b[0m         \u001b[0mprocessed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcolor_process\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 177\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_model\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mprocessed\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbatch_size\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    178\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    179\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mpredict_one\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/src/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     63\u001b[0m         \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     64\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 65\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     66\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     67\u001b[0m             \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/src/engine/training.py\u001b[0m in \u001b[0;36mpredict\u001b[0;34m(self, x, batch_size, verbose, steps, callbacks, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   2653\u001b[0m                     \u001b[0;32mfor\u001b[0m \u001b[0mstep\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msteps\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2654\u001b[0m                         \u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_predict_batch_begin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mstep\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2655\u001b[0;31m                         \u001b[0mtmp_batch_outputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0miterator\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   2656\u001b[0m                         \u001b[0;32mif\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshould_sync\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2657\u001b[0m                             \u001b[0mcontext\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masync_wait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/util/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    148\u001b[0m     \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    149\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 150\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    151\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    152\u001b[0m       \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    830\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    831\u001b[0m       \u001b[0;32mwith\u001b[0m \u001b[0mOptionalXlaContext\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_jit_compile\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 832\u001b[0;31m         \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    833\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    834\u001b[0m       \u001b[0mnew_tracing_count\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexperimental_get_tracing_count\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py\u001b[0m in \u001b[0;36m_call\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    875\u001b[0m       \u001b[0;31m# In this case we have not created variables on the first call. So we can\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    876\u001b[0m       \u001b[0;31m# run the first trace but we should fail if variables are created.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 877\u001b[0;31m       results = tracing_compilation.call_function(\n\u001b[0m\u001b[1;32m    878\u001b[0m           \u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkwds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_variable_creation_config\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    879\u001b[0m       )\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py\u001b[0m in \u001b[0;36mcall_function\u001b[0;34m(args, kwargs, tracing_options)\u001b[0m\n\u001b[1;32m    137\u001b[0m   \u001b[0mbound_args\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunction\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfunction_type\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbind\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    138\u001b[0m   \u001b[0mflat_inputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunction\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfunction_type\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0munpack_inputs\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbound_args\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 139\u001b[0;31m   return function._call_flat(  # pylint: disable=protected-access\n\u001b[0m\u001b[1;32m    140\u001b[0m       \u001b[0mflat_inputs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcaptured_inputs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mfunction\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcaptured_inputs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    141\u001b[0m   )\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/eager/polymorphic_function/concrete_function.py\u001b[0m in \u001b[0;36m_call_flat\u001b[0;34m(self, tensor_inputs, captured_inputs)\u001b[0m\n\u001b[1;32m   1321\u001b[0m         and executing_eagerly):\n\u001b[1;32m   1322\u001b[0m       \u001b[0;31m# No tape is watching; skip to running the function.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1323\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_inference_function\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcall_preflattened\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1324\u001b[0m     forward_backward = self._select_forward_and_backward_functions(\n\u001b[1;32m   1325\u001b[0m         \u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/eager/polymorphic_function/atomic_function.py\u001b[0m in \u001b[0;36mcall_preflattened\u001b[0;34m(self, args)\u001b[0m\n\u001b[1;32m    214\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0mcall_preflattened\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0margs\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mSequence\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mcore\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTensor\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0mAny\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    215\u001b[0m     \u001b[0;34m\"\"\"Calls with flattened tensor inputs and returns the structured output.\"\"\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 216\u001b[0;31m     \u001b[0mflat_outputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcall_flat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    217\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfunction_type\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpack_output\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mflat_outputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    218\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/eager/polymorphic_function/atomic_function.py\u001b[0m in \u001b[0;36mcall_flat\u001b[0;34m(self, *args)\u001b[0m\n\u001b[1;32m    249\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mrecord\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstop_recording\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    250\u001b[0m           \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_bound_context\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexecuting_eagerly\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 251\u001b[0;31m             outputs = self._bound_context.call_function(\n\u001b[0m\u001b[1;32m    252\u001b[0m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mname\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    253\u001b[0m                 \u001b[0mlist\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/eager/context.py\u001b[0m in \u001b[0;36mcall_function\u001b[0;34m(self, name, tensor_inputs, num_outputs)\u001b[0m\n\u001b[1;32m   1484\u001b[0m     \u001b[0mcancellation_context\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcancellation\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcontext\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1485\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mcancellation_context\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1486\u001b[0;31m       outputs = execute.execute(\n\u001b[0m\u001b[1;32m   1487\u001b[0m           \u001b[0mname\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdecode\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"utf-8\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1488\u001b[0m           \u001b[0mnum_outputs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mnum_outputs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/eager/execute.py\u001b[0m in \u001b[0;36mquick_execute\u001b[0;34m(op_name, num_outputs, inputs, attrs, ctx, name)\u001b[0m\n\u001b[1;32m     51\u001b[0m   \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     52\u001b[0m     \u001b[0mctx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mensure_initialized\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 53\u001b[0;31m     tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,\n\u001b[0m\u001b[1;32m     54\u001b[0m                                         inputs, attrs, num_outputs)\n\u001b[1;32m     55\u001b[0m   \u001b[0;32mexcept\u001b[0m \u001b[0mcore\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_NotOkStatusException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ],
      "source": [
        "untargeted = attack_all(models, samples=100, targeted=False)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 42,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "WorYXy6P-3zP",
        "outputId": "cb70fbd0-84f4-4968-dbee-1457b299acb0"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            " resnet - image 8416 - 1 / 10\n",
            "Attacking with target airplane\n",
            "4/4 [==============================] - 3s 671ms/step\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/content/drive/MyDrive/one-pixel-attack-keras-master/differential_evolution.py:159: RuntimeWarning: divide by zero encountered in scalar divide\n",
            "  if self.callback and self.callback(self._scale_parameters(self.population[0]), convergence=self.tol / convergence):\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 39ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 669ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 390ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 546ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 690ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 379ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 665ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 609ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 42ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 675ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 678ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 394ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 390ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 450ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 689ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 36ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 442ms/step\n",
            "1/1 [==============================] - 0s 48ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 608ms/step\n",
            "1/1 [==============================] - 0s 37ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 556ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 656ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 379ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 57ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 678ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 396ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 687ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 46ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 687ms/step\n",
            "1/1 [==============================] - 0s 45ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 437ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 627ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 410ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 667ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 43ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 655ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 36ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 538ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 643ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Attacking with target automobile\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 690ms/step\n",
            "1/1 [==============================] - 0s 42ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 531ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 395ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 643ms/step\n",
            "1/1 [==============================] - 0s 46ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 648ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 377ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 695ms/step\n",
            "1/1 [==============================] - 0s 46ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 556ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 490ms/step\n",
            "1/1 [==============================] - 0s 55ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 690ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 37ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 390ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 50ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 608ms/step\n",
            "1/1 [==============================] - 0s 36ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 683ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 390ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 542ms/step\n",
            "1/1 [==============================] - 0s 52ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 694ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 465ms/step\n",
            "1/1 [==============================] - 0s 48ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 642ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 52ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 694ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 391ms/step\n",
            "1/1 [==============================] - 0s 36ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 443ms/step\n",
            "1/1 [==============================] - 0s 42ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 643ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 37ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 604ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 622ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 400ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 379ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 663ms/step\n",
            "1/1 [==============================] - 0s 46ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 460ms/step\n",
            "1/1 [==============================] - 0s 39ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 379ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 379ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 689ms/step\n",
            "1/1 [==============================] - 0s 48ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 532ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 393ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 50ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 669ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 396ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Attacking with target bird\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "4/4 [==============================] - 2s 396ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 649ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 394ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 533ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 672ms/step\n",
            "1/1 [==============================] - 0s 52ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 652ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 637ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 392ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 394ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 627ms/step\n",
            "1/1 [==============================] - 0s 44ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 564ms/step\n",
            "1/1 [==============================] - 0s 43ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 390ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 676ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 389ms/step\n",
            "1/1 [==============================] - 0s 50ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 690ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 381ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 405ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 679ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 379ms/step\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 380ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 573ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 639ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 679ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 38ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 39ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 676ms/step\n",
            "1/1 [==============================] - 0s 54ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 447ms/step\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 395ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 648ms/step\n",
            "1/1 [==============================] - 0s 46ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 631ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 393ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 390ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 384ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 684ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 383ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 388ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 390ms/step\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 382ms/step\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 664ms/step\n",
            "1/1 [==============================] - 0s 47ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 472ms/step\n",
            "1/1 [==============================] - 0s 50ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 386ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 418ms/step\n",
            "1/1 [==============================] - 0s 49ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 3s 675ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 385ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 387ms/step\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "Predicted Class: frog\n",
            "4/4 [==============================] - 2s 395ms/step\n"
          ]
        },
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-42-0bd55d8b9b66>\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mtargeted\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mattack_all\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodels\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msamples\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m10\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtargeted\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;32m<ipython-input-40-67afb077e253>\u001b[0m in \u001b[0;36mattack_all\u001b[0;34m(models, samples, pixels, targeted, maxiter, popsize, verbose)\u001b[0m\n\u001b[1;32m     17\u001b[0m                         \u001b[0;32mif\u001b[0m \u001b[0mtarget\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0my_test\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mimg_id\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m                             \u001b[0;32mcontinue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 19\u001b[0;31m                     result = attack(img_id, model, target, pixel_count, \n\u001b[0m\u001b[1;32m     20\u001b[0m                                     \u001b[0mmaxiter\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mmaxiter\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpopsize\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mpopsize\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     21\u001b[0m                                     verbose=verbose)\n",
            "\u001b[0;32m<ipython-input-34-a3ba829b94fd>\u001b[0m in \u001b[0;36mattack\u001b[0;34m(img_id, model, target, pixel_count, maxiter, popsize, verbose)\u001b[0m\n\u001b[1;32m     22\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     23\u001b[0m     \u001b[0;31m# Call Scipy's Implementation of Differential Evolution\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 24\u001b[0;31m     attack_result = differential_evolution(\n\u001b[0m\u001b[1;32m     25\u001b[0m         \u001b[0mpredict_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbounds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmaxiter\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mmaxiter\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpopsize\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mpopmul\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     26\u001b[0m         recombination=1, atol=-1, callback=callback_fn, polish=False)\n",
            "\u001b[0;32m/content/drive/MyDrive/one-pixel-attack-keras-master/differential_evolution.py\u001b[0m in \u001b[0;36mdifferential_evolution\u001b[0;34m(func, bounds, args, strategy, maxiter, popsize, tol, mutation, recombination, seed, callback, disp, polish, init, atol)\u001b[0m\n\u001b[1;32m     20\u001b[0m                                          \u001b[0mcallback\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcallback\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     21\u001b[0m                                          disp=disp, init=init, atol=atol)\n\u001b[0;32m---> 22\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0msolver\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msolve\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     23\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     24\u001b[0m \u001b[0;32mclass\u001b[0m \u001b[0mDifferentialEvolutionSolver\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/content/drive/MyDrive/one-pixel-attack-keras-master/differential_evolution.py\u001b[0m in \u001b[0;36msolve\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    146\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mnit\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmaxiter\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    147\u001b[0m             \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 148\u001b[0;31m                 \u001b[0mnext\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    149\u001b[0m             \u001b[0;32mexcept\u001b[0m \u001b[0mStopIteration\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    150\u001b[0m                 \u001b[0mwarning_flag\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/content/drive/MyDrive/one-pixel-attack-keras-master/differential_evolution.py\u001b[0m in \u001b[0;36m__next__\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    219\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mtrial\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mtrials\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_ensure_constraint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrial\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    220\u001b[0m         \u001b[0mparameters\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_scale_parameters\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrial\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mtrial\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mtrials\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 221\u001b[0;31m         \u001b[0menergies\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mparameters\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    222\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_nfev\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0mitersize\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    223\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-34-a3ba829b94fd>\u001b[0m in \u001b[0;36mpredict_fn\u001b[0;34m(xs)\u001b[0m\n\u001b[1;32m     14\u001b[0m     \u001b[0;31m# Format the predict/callback functions for the differential evolution algorithm\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mpredict_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mxs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 16\u001b[0;31m         return predict_classes(xs, x_test[img_id], target_class, \n\u001b[0m\u001b[1;32m     17\u001b[0m                                model, target is None)\n\u001b[1;32m     18\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-20-713e560dbc26>\u001b[0m in \u001b[0;36mpredict_classes\u001b[0;34m(xs, img, target_class, model, minimize)\u001b[0m\n\u001b[1;32m      2\u001b[0m     \u001b[0;31m# Perturb the image with the given pixel(s) x and get the prediction of the model\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m     \u001b[0mimgs_perturbed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mperturb_image\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mxs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 4\u001b[0;31m     \u001b[0mpredictions\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mimgs_perturbed\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtarget_class\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      5\u001b[0m     \u001b[0;31m# This function should always be minimized, so return its complement if needed\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mpredictions\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mminimize\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0;36m1\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mpredictions\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/content/drive/MyDrive/one-pixel-attack-keras-master/resnet.py\u001b[0m in \u001b[0;36mpredict\u001b[0;34m(self, img)\u001b[0m\n\u001b[1;32m    175\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    176\u001b[0m         \u001b[0mprocessed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcolor_process\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 177\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_model\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mprocessed\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbatch_size\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    178\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    179\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mpredict_one\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/src/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     63\u001b[0m         \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     64\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 65\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     66\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     67\u001b[0m             \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ],
      "source": [
        "targeted = attack_all(models, samples=10, targeted=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yU3xYE9A-3zP"
      },
      "source": [
        "### Attack Statistics"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QgnneS3u-3zP"
      },
      "source": [
        "Print the final results!"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "4TVQJqS5-3zP"
      },
      "outputs": [],
      "source": [
        "# Load the results\n",
        "untargeted, targeted = helper.load_results()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "scrolled": true,
        "id": "KTd2abmO-3zP"
      },
      "outputs": [],
      "source": [
        "columns = ['model', 'pixels', 'image', 'true', 'predicted', 'success', 'cdiff', 'prior_probs', 'predicted_probs', 'perturbation']\n",
        "untargeted_results = pd.DataFrame(untargeted, columns=columns)\n",
        "targeted_results = pd.DataFrame(targeted, columns=columns)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kt23jsn_-3zP"
      },
      "source": [
        "#### Untargeted"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "b8XR3WOV-3zP",
        "outputId": "c16ceb41-0f82-454d-d7f1-b9f252647d7d"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "<div>\n",
              "<style>\n",
              "    .dataframe thead tr:only-child th {\n",
              "        text-align: right;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: left;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>model</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>pixels</th>\n",
              "      <th>attack_success_rate</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>lenet</td>\n",
              "      <td>0.7488</td>\n",
              "      <td>1</td>\n",
              "      <td>0.63</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>lenet</td>\n",
              "      <td>0.7488</td>\n",
              "      <td>3</td>\n",
              "      <td>0.92</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>lenet</td>\n",
              "      <td>0.7488</td>\n",
              "      <td>5</td>\n",
              "      <td>0.93</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>pure_cnn</td>\n",
              "      <td>0.8877</td>\n",
              "      <td>1</td>\n",
              "      <td>0.13</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>pure_cnn</td>\n",
              "      <td>0.8877</td>\n",
              "      <td>3</td>\n",
              "      <td>0.58</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>pure_cnn</td>\n",
              "      <td>0.8877</td>\n",
              "      <td>5</td>\n",
              "      <td>0.63</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>net_in_net</td>\n",
              "      <td>0.9074</td>\n",
              "      <td>1</td>\n",
              "      <td>0.34</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>net_in_net</td>\n",
              "      <td>0.9074</td>\n",
              "      <td>3</td>\n",
              "      <td>0.73</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>net_in_net</td>\n",
              "      <td>0.9074</td>\n",
              "      <td>5</td>\n",
              "      <td>0.73</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>resnet</td>\n",
              "      <td>0.9231</td>\n",
              "      <td>1</td>\n",
              "      <td>0.34</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>resnet</td>\n",
              "      <td>0.9231</td>\n",
              "      <td>3</td>\n",
              "      <td>0.79</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>resnet</td>\n",
              "      <td>0.9231</td>\n",
              "      <td>5</td>\n",
              "      <td>0.79</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>densenet</td>\n",
              "      <td>0.9467</td>\n",
              "      <td>1</td>\n",
              "      <td>0.31</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>densenet</td>\n",
              "      <td>0.9467</td>\n",
              "      <td>3</td>\n",
              "      <td>0.71</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>densenet</td>\n",
              "      <td>0.9467</td>\n",
              "      <td>5</td>\n",
              "      <td>0.69</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15</th>\n",
              "      <td>wide_resnet</td>\n",
              "      <td>0.9534</td>\n",
              "      <td>1</td>\n",
              "      <td>0.19</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16</th>\n",
              "      <td>wide_resnet</td>\n",
              "      <td>0.9534</td>\n",
              "      <td>3</td>\n",
              "      <td>0.58</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17</th>\n",
              "      <td>wide_resnet</td>\n",
              "      <td>0.9534</td>\n",
              "      <td>5</td>\n",
              "      <td>0.65</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18</th>\n",
              "      <td>capsnet</td>\n",
              "      <td>0.7982</td>\n",
              "      <td>1</td>\n",
              "      <td>0.19</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>capsnet</td>\n",
              "      <td>0.7982</td>\n",
              "      <td>3</td>\n",
              "      <td>0.39</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>20</th>\n",
              "      <td>capsnet</td>\n",
              "      <td>0.7982</td>\n",
              "      <td>5</td>\n",
              "      <td>0.36</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "          model  accuracy  pixels  attack_success_rate\n",
              "0     lenet    0.7488       1                 0.63\n",
              "1     lenet    0.7488       3                 0.92\n",
              "2     lenet    0.7488       5                 0.93\n",
              "3      pure_cnn    0.8877       1                 0.13\n",
              "4      pure_cnn    0.8877       3                 0.58\n",
              "5      pure_cnn    0.8877       5                 0.63\n",
              "6    net_in_net    0.9074       1                 0.34\n",
              "7    net_in_net    0.9074       3                 0.73\n",
              "8    net_in_net    0.9074       5                 0.73\n",
              "9        resnet    0.9231       1                 0.34\n",
              "10       resnet    0.9231       3                 0.79\n",
              "11       resnet    0.9231       5                 0.79\n",
              "12     densenet    0.9467       1                 0.31\n",
              "13     densenet    0.9467       3                 0.71\n",
              "14     densenet    0.9467       5                 0.69\n",
              "15  wide_resnet    0.9534       1                 0.19\n",
              "16  wide_resnet    0.9534       3                 0.58\n",
              "17  wide_resnet    0.9534       5                 0.65\n",
              "18      capsnet    0.7982       1                 0.19\n",
              "19      capsnet    0.7982       3                 0.39\n",
              "20      capsnet    0.7982       5                 0.36"
            ]
          },
          "execution_count": 21,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "helper.attack_stats(untargeted_results, models, network_stats)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "g58JTXjK-3zP"
      },
      "source": [
        "#### Targeted"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "SZ62dQEs-3zP",
        "outputId": "35843b9a-9ddc-478e-a3d3-86c44f7197e1"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "<div>\n",
              "<style>\n",
              "    .dataframe thead tr:only-child th {\n",
              "        text-align: right;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: left;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>model</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>pixels</th>\n",
              "      <th>attack_success_rate</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>lenet</td>\n",
              "      <td>0.7488</td>\n",
              "      <td>1</td>\n",
              "      <td>0.344444</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>lenet</td>\n",
              "      <td>0.7488</td>\n",
              "      <td>3</td>\n",
              "      <td>0.644444</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>lenet</td>\n",
              "      <td>0.7488</td>\n",
              "      <td>5</td>\n",
              "      <td>0.644444</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>pure_cnn</td>\n",
              "      <td>0.8877</td>\n",
              "      <td>1</td>\n",
              "      <td>0.066667</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>pure_cnn</td>\n",
              "      <td>0.8877</td>\n",
              "      <td>3</td>\n",
              "      <td>0.133333</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>pure_cnn</td>\n",
              "      <td>0.8877</td>\n",
              "      <td>5</td>\n",
              "      <td>0.188889</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>net_in_net</td>\n",
              "      <td>0.9074</td>\n",
              "      <td>1</td>\n",
              "      <td>0.100000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>net_in_net</td>\n",
              "      <td>0.9074</td>\n",
              "      <td>3</td>\n",
              "      <td>0.244444</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>net_in_net</td>\n",
              "      <td>0.9074</td>\n",
              "      <td>5</td>\n",
              "      <td>0.311111</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>resnet</td>\n",
              "      <td>0.9231</td>\n",
              "      <td>1</td>\n",
              "      <td>0.144444</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>resnet</td>\n",
              "      <td>0.9231</td>\n",
              "      <td>3</td>\n",
              "      <td>0.211111</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>resnet</td>\n",
              "      <td>0.9231</td>\n",
              "      <td>5</td>\n",
              "      <td>0.222222</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>densenet</td>\n",
              "      <td>0.9467</td>\n",
              "      <td>1</td>\n",
              "      <td>0.044444</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>densenet</td>\n",
              "      <td>0.9467</td>\n",
              "      <td>3</td>\n",
              "      <td>0.233333</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>densenet</td>\n",
              "      <td>0.9467</td>\n",
              "      <td>5</td>\n",
              "      <td>0.288889</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15</th>\n",
              "      <td>wide_resnet</td>\n",
              "      <td>0.9534</td>\n",
              "      <td>1</td>\n",
              "      <td>0.011111</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16</th>\n",
              "      <td>wide_resnet</td>\n",
              "      <td>0.9534</td>\n",
              "      <td>3</td>\n",
              "      <td>0.188889</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17</th>\n",
              "      <td>wide_resnet</td>\n",
              "      <td>0.9534</td>\n",
              "      <td>5</td>\n",
              "      <td>0.222222</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18</th>\n",
              "      <td>capsnet</td>\n",
              "      <td>0.7982</td>\n",
              "      <td>1</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>capsnet</td>\n",
              "      <td>0.7982</td>\n",
              "      <td>3</td>\n",
              "      <td>0.044444</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>20</th>\n",
              "      <td>capsnet</td>\n",
              "      <td>0.7982</td>\n",
              "      <td>5</td>\n",
              "      <td>0.044444</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "          model  accuracy  pixels  attack_success_rate\n",
              "0     lenet    0.7488       1             0.344444\n",
              "1     lenet    0.7488       3             0.644444\n",
              "2     lenet    0.7488       5             0.644444\n",
              "3      pure_cnn    0.8877       1             0.066667\n",
              "4      pure_cnn    0.8877       3             0.133333\n",
              "5      pure_cnn    0.8877       5             0.188889\n",
              "6    net_in_net    0.9074       1             0.100000\n",
              "7    net_in_net    0.9074       3             0.244444\n",
              "8    net_in_net    0.9074       5             0.311111\n",
              "9        resnet    0.9231       1             0.144444\n",
              "10       resnet    0.9231       3             0.211111\n",
              "11       resnet    0.9231       5             0.222222\n",
              "12     densenet    0.9467       1             0.044444\n",
              "13     densenet    0.9467       3             0.233333\n",
              "14     densenet    0.9467       5             0.288889\n",
              "15  wide_resnet    0.9534       1             0.011111\n",
              "16  wide_resnet    0.9534       3             0.188889\n",
              "17  wide_resnet    0.9534       5             0.222222\n",
              "18      capsnet    0.7982       1             0.000000\n",
              "19      capsnet    0.7982       3             0.044444\n",
              "20      capsnet    0.7982       5             0.044444"
            ]
          },
          "execution_count": 22,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "helper.attack_stats(targeted_results, models, network_stats)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rY4A_54b-3zP"
      },
      "source": [
        "### Show some successful attacks"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mqV_dqfE-3zP"
      },
      "source": [
        "Plot 9 random successful attack images"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "IU3nIzQ4-3zQ",
        "outputId": "ae2184b5-74e6-4c19-8191-fd377658e36e"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Untargeted Attack\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f91b1546b38>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "print('Untargeted Attack')\n",
        "helper.visualize_attack(untargeted_results, class_names)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Tzynlf1S-3zQ",
        "outputId": "7e209cea-1e5d-4d8f-886e-21282251fa89"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Targeted Attack\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f91b9546780>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "print('Targeted Attack')\n",
        "helper.visualize_attack(targeted_results, class_names)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PKAc_Xk2-3zQ"
      },
      "source": [
        "## Conclusions"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dBJoL0f8-3zQ"
      },
      "source": [
        "It appears that the accuracy of a model is not strongly correlated with the chance of performing a successful attack on an image. Perhaps surprisingly, the purely convolutional model is the most resistant CNN to these types of attacks. In addition, the capsule network CapsNet has the lowest attack success rate out of all the models, although it is still vulnerable to attack.\n",
        "\n",
        "[Part 2](1_one-pixel-attack-cifar10.ipynb)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3ZSiG_eJ-3zQ"
      },
      "source": [
        "## Credits\n",
        " -  This implemenation is based off of the original paper describing the one pixel attack: https://arxiv.org/abs/1710.08864\n",
        " - Base code for iPython notebook: https://github.com/09rohanchopra/cifar10\n",
        " - Keras Cifar10 models: https://github.com/BIGBALLON/cifar-10-cnn\n",
        " - Scipy's differential evolution implementation: https://docs.scipy.org/doc/scipy-0.17.0/reference/generated/scipy.optimize.differential_evolution.html\n",
        " - State of the art: https://github.com/RedditSota/state-of-the-art-result-for-machine-learning-problems\n",
        " - CapsNet Keras: https://github.com/XifengGuo/CapsNet-Keras\n",
        " - CapsNet with Cifar: https://github.com/theblackcat102/dynamic-routing-capsule-cifar"
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3 (ipykernel)",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.11.5"
    },
    "colab": {
      "provenance": []
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}