BurmeseHandwrittenDigit.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "BHDD-Dataset.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "JojXy1zDcAW7",
        "m_x_8tOVLJq7",
        "4icEIrjGnsju"
      ],
      "authorship_tag": "ABX9TyMdwt4cG06EWLKXEaUOD2Zv",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/ThuraAung1601/BHDD-using-streamlit/blob/main/BurmeseHandwrittenDigit.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1QziLZUcLO-H"
      },
      "source": [
        "# BHDD Dataset Preprocessing \n",
        "- Dataset Link : https://github.com/baseresearch/BHDD\n",
        "( Download raw file directly and upload to your drive for further uses )"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hgknVq80BF5_",
        "outputId": "9847e54b-6d95-451a-fcf0-414ceccabc9b"
      },
      "source": [
        "from google.colab import drive\n",
        "\n",
        "drive.mount('/content/gdrive')"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/gdrive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mA-pP8VQtrM0"
      },
      "source": [
        ""
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JojXy1zDcAW7"
      },
      "source": [
        "## Data Loading"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oP3h4gjRBfg8",
        "outputId": "d8ea6110-5fff-4345-ed2d-ed8c7c2f6df9"
      },
      "source": [
        "import pickle\n",
        "import numpy as np\n",
        "\n",
        "#url_path : Path of dataset in your drive\n",
        "url_path = \"/content/gdrive/MyDrive/BHDD Experiments /data.pkl\" \n",
        "\n",
        "dataset = []\n",
        "\n",
        "with open(url_path,\"rb\") as file:\n",
        "    dataset = pickle.load(file)\n",
        "    \n",
        "trainDataset = dataset[\"trainDataset\"]\n",
        "testDataset = dataset[\"testDataset\"]\n",
        "\n",
        "print(\"Train Size: {} Test Size: {}\".format(len(trainDataset), len(testDataset)))\n",
        "            "
      ],
      "execution_count": 76,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Train Size: 60000 Test Size: 27561\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fFvTctK0XMPi"
      },
      "source": [
        "import urllib.request\n",
        "import gzip\n",
        "import pickle\n",
        "import os\n",
        "import numpy as np\n",
        "from PIL import Image"
      ],
      "execution_count": 77,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "chd3bMV1ICJH"
      },
      "source": [
        "numberOfClass = 10\n",
        "\n",
        "trainX = [list(td[\"image\"] / 255) for td in trainDataset]\n",
        "trainY = np.eye(numberOfClass)[np.array([td[\"label\"] for td in trainDataset]).reshape(-1)].tolist() #One-hot encode training labels\n",
        "testX = [list(td[\"image\"] / 255) for td in testDataset]\n",
        "testY = np.eye(numberOfClass)[np.array([td[\"label\"] for td in testDataset]).reshape(-1)].tolist() #One-hot encode test labels"
      ],
      "execution_count": 78,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "t46m-NSZIdd1",
        "outputId": "3a694ec1-101d-4f19-e178-3f81043d0e31"
      },
      "source": [
        "print(\"TrainImageShape:\" + str(np.shape(trainX)),\n",
        "     \"TrainLabelShape:\" + str(np.shape(trainY)),\n",
        "     \"TestImageShape:\" + str(np.shape(testX)),\n",
        "     \"TestLabelShape:\" + str(np.shape(testY)))"
      ],
      "execution_count": 79,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "TrainImageShape:(60000, 28, 28) TrainLabelShape:(60000, 10) TestImageShape:(27561, 28, 28) TestLabelShape:(27561, 10)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 287
        },
        "id": "RZc758QlQ8fs",
        "outputId": "1ed0cf15-f52a-4fbc-d6c7-ededfa0c8e17"
      },
      "source": [
        "plt.imshow(trainX[6],cmap='gray')"
      ],
      "execution_count": 98,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f2c66b362d0>"
            ]
          },
          "metadata": {},
          "execution_count": 98
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rojO3MostwTr"
      },
      "source": [
        "# Deep Learning Algorithms\n",
        "- Perceptron\n",
        "- MLP\n",
        "- ConvNet"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "m_x_8tOVLJq7"
      },
      "source": [
        "## Visualize the Dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xmq22SGsIf2z",
        "outputId": "ed7ce1b4-e577-4ec2-9686-02b43c63b2d5"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "%pylab inline\n",
        "\n",
        "def show_images(x, y, title_str):\n",
        "    for c in range(1, 10):\n",
        "        subplot(3, 3, c)\n",
        "        i = randint(len(x)) \n",
        "        im = np.reshape(x[i], (28,28))\n",
        "        axis(\"off\")\n",
        "        label = np.argmax(y[i]) \n",
        "        title(\"{} = {}\".format(title_str, label))\n",
        "        imshow(im, cmap='Greys')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Populating the interactive namespace from numpy and matplotlib\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 264
        },
        "id": "YfnhPcWiIlBt",
        "outputId": "8759ffd1-b043-4da9-c6c9-daa33a6de690"
      },
      "source": [
        "### Display Training set\n",
        "\n",
        "show_images(trainX, trainY, 'Label')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 9 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 264
        },
        "id": "ixxCfPtrIsoP",
        "outputId": "99212c84-c94c-48e6-9d89-251df3366b3e"
      },
      "source": [
        "### Display Test set\n",
        "\n",
        "show_images(testX, testY, 'Label')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 9 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Eb3f7Wm0YE4T"
      },
      "source": [
        "## Perceptron"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1gE6xjg5ZLN5"
      },
      "source": [
        "import numpy as np\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BpzQ-poIcYe1",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "47d30057-a654-43f8-b58f-d8a98265250f"
      },
      "source": [
        "num_classes = 10\n",
        "input_shape = (28, 28, 1)\n",
        "model = keras.Sequential(\n",
        "    [\n",
        "        keras.Input(shape=input_shape),\n",
        "        layers.Flatten(),\n",
        "        layers.Dense(num_classes, activation=\"softmax\"),\n",
        "    ]\n",
        ")\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "flatten (Flatten)            (None, 784)               0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 10)                7850      \n",
            "=================================================================\n",
            "Total params: 7,850\n",
            "Trainable params: 7,850\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QelmkQr9cfnd",
        "outputId": "04f782ea-94e8-4966-d966-7df0df7c3b8b"
      },
      "source": [
        "# Scale images to the [0, 1] range\n",
        "trainX = np.array(trainX)\n",
        "testX = np.array(testX)\n",
        "\n",
        "#Normalize \n",
        "trainX = trainX.astype(\"float32\") / 255\n",
        "testX = testX.astype(\"float32\") / 255\n",
        "\n",
        "# Make sure images have shape (28, 28, 1)\n",
        "trainX = np.expand_dims(trainX, -1)\n",
        "testX = np.expand_dims(testX, -1)\n",
        "\n",
        "print(\"x_train shape:\", trainX.shape)\n",
        "print(trainX.shape[0], \"train samples\")\n",
        "print(testX.shape[0], \"test samples\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "x_train shape: (60000, 28, 28, 1)\n",
            "60000 train samples\n",
            "27561 test samples\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uWuPprMmWjWJ"
      },
      "source": [
        "trainY = np.array(trainY)\n",
        "testY = np.array(testY)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AyS8aSjhdAlH"
      },
      "source": [
        "batch_size = 128\n",
        "epochs = 15\n",
        "\n",
        "model.compile(loss=\"categorical_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\",keras.metrics.Precision(),keras.metrics.Recall()])\n",
        "\n",
        "history = model.fit(trainX, trainY, batch_size=batch_size, epochs=epochs, validation_split=0.1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b7XwDce9dOgd",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "0a08adf7-8888-4bba-e65f-60b7ac8cf2c8"
      },
      "source": [
        "print(history.history.keys())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "dict_keys(['loss', 'accuracy', 'precision', 'recall', 'val_loss', 'val_accuracy', 'val_precision', 'val_recall'])\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PJqDF-8UaJaF",
        "outputId": "b8f3debf-4758-42fc-fd4d-5e7b80df1d1a"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title(\"model accuracy\")\n",
        "plt.ylabel(\"accuracy\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.legend(['train','test'],loc='lower right')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "2SZBgAKFahgv",
        "outputId": "d616d2fd-4ae4-4e6a-80b2-9f8da40026a6"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title(\"model loss\")\n",
        "plt.ylabel(\"loss\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.legend(['train','test'],loc='upper right')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "aLmiAU27qLrm",
        "outputId": "ac738b89-2b97-4a15-b83a-8634c0255825"
      },
      "source": [
        "score = model.evaluate(testX, testY, verbose=0)\n",
        "print(\"Test loss:\", score[0])\n",
        "print(\"Test accuracy:\", score[1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Test loss: 1.6300876140594482\n",
            "Test accuracy: 0.8476107716560364\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xp5MQvUsClxv",
        "outputId": "d66e5e70-1523-484b-c2dd-662b42c62502"
      },
      "source": [
        "loss, accuracy, precision, recall = model.evaluate(testX, testY, verbose=0)\n",
        "print('Loss', loss)\n",
        "print('Accuracy', accuracy)\n",
        "print('Precision', precision)\n",
        "print('Recall', recall)\n",
        "\n",
        "f1_score = (2* recall*precision)/(recall+precision)\n",
        "print(\"f1_score\", f1_score)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Loss 1.6300876140594482\n",
            "Accuracy 0.8476107716560364\n",
            "Precision 1.0\n",
            "Recall 0.0009070788510143757\n",
            "f1_score 0.0018125136092666098\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GcdKkqfKbVmM"
      },
      "source": [
        "## Multilayer Perceptron "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7cBpOfnObX3X",
        "outputId": "48828344-adbc-4a64-de29-c00a932e5dcb"
      },
      "source": [
        "num_classes = 10\n",
        "input_shape = (28, 28, 1)\n",
        "model = keras.Sequential(\n",
        "    [\n",
        "        keras.Input(shape=input_shape),\n",
        "        layers.Flatten(),\n",
        "        layers.Dense(64, activation=\"relu\"),\n",
        "        layers.Dense(num_classes, activation=\"softmax\"),\n",
        "    ]\n",
        ")\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 64,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_5\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "flatten_5 (Flatten)          (None, 784)               0         \n",
            "_________________________________________________________________\n",
            "dense_14 (Dense)             (None, 64)                50240     \n",
            "_________________________________________________________________\n",
            "dense_15 (Dense)             (None, 10)                650       \n",
            "=================================================================\n",
            "Total params: 50,890\n",
            "Trainable params: 50,890\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "l_Op0PmobX3c",
        "outputId": "122ccdd7-10a0-4e21-fef4-77f78fde79dc"
      },
      "source": [
        "# Scale images to the [0, 1] range\n",
        "trainX = np.array(trainX)\n",
        "testX = np.array(testX)\n",
        "trainX = trainX.astype(\"float32\") / 255\n",
        "testX = testX.astype(\"float32\") / 255\n",
        "\n",
        "# Make sure images have shape (28, 28, 1)\n",
        "trainX = np.expand_dims(trainX, -1)\n",
        "testX = np.expand_dims(testX, -1)\n",
        "\n",
        "print(\"x_train shape:\", trainX.shape)\n",
        "print(trainX.shape[0], \"train samples\")\n",
        "print(testX.shape[0], \"test samples\")"
      ],
      "execution_count": 65,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "x_train shape: (60000, 28, 28, 1)\n",
            "60000 train samples\n",
            "27561 test samples\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Z1qrsbQSbX3d"
      },
      "source": [
        "trainY = np.array(trainY)\n",
        "testY = np.array(testY)"
      ],
      "execution_count": 66,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "49jU1m2pbX3e",
        "outputId": "dd8c8fec-4920-45c4-ac80-413ea1d5e113"
      },
      "source": [
        "batch_size = 128\n",
        "epochs = 15\n",
        "\n",
        "model.compile(loss=\"categorical_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\",keras.metrics.Precision(),keras.metrics.Recall()])\n",
        "\n",
        "history = model.fit(trainX, trainY, batch_size=batch_size, epochs=epochs, validation_split=0.1)"
      ],
      "execution_count": 67,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/15\n",
            "422/422 [==============================] - 3s 5ms/step - loss: 2.0900 - accuracy: 0.5695 - precision_5: 0.0000e+00 - recall_5: 0.0000e+00 - val_loss: 1.7588 - val_accuracy: 0.7505 - val_precision_5: 0.0000e+00 - val_recall_5: 0.0000e+00\n",
            "Epoch 2/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 1.4432 - accuracy: 0.7648 - precision_5: 0.9951 - recall_5: 0.0411 - val_loss: 1.1820 - val_accuracy: 0.7917 - val_precision_5: 0.9936 - val_recall_5: 0.1295\n",
            "Epoch 3/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.9809 - accuracy: 0.8369 - precision_5: 0.9829 - recall_5: 0.3031 - val_loss: 0.8254 - val_accuracy: 0.8578 - val_precision_5: 0.9815 - val_recall_5: 0.4522\n",
            "Epoch 4/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.6964 - accuracy: 0.8841 - precision_5: 0.9777 - recall_5: 0.5888 - val_loss: 0.6036 - val_accuracy: 0.8943 - val_precision_5: 0.9747 - val_recall_5: 0.6883\n",
            "Epoch 5/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.5159 - accuracy: 0.9104 - precision_5: 0.9737 - recall_5: 0.7654 - val_loss: 0.4617 - val_accuracy: 0.9180 - val_precision_5: 0.9713 - val_recall_5: 0.8052\n",
            "Epoch 6/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.4000 - accuracy: 0.9264 - precision_5: 0.9715 - recall_5: 0.8453 - val_loss: 0.3690 - val_accuracy: 0.9293 - val_precision_5: 0.9697 - val_recall_5: 0.8652\n",
            "Epoch 7/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.3240 - accuracy: 0.9379 - precision_5: 0.9702 - recall_5: 0.8869 - val_loss: 0.3080 - val_accuracy: 0.9378 - val_precision_5: 0.9698 - val_recall_5: 0.8942\n",
            "Epoch 8/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.2724 - accuracy: 0.9448 - precision_5: 0.9702 - recall_5: 0.9101 - val_loss: 0.2661 - val_accuracy: 0.9437 - val_precision_5: 0.9695 - val_recall_5: 0.9115\n",
            "Epoch 9/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.2355 - accuracy: 0.9510 - precision_5: 0.9717 - recall_5: 0.9250 - val_loss: 0.2352 - val_accuracy: 0.9452 - val_precision_5: 0.9683 - val_recall_5: 0.9212\n",
            "Epoch 10/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.2082 - accuracy: 0.9557 - precision_5: 0.9724 - recall_5: 0.9350 - val_loss: 0.2113 - val_accuracy: 0.9535 - val_precision_5: 0.9719 - val_recall_5: 0.9347\n",
            "Epoch 11/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.1875 - accuracy: 0.9591 - precision_5: 0.9735 - recall_5: 0.9425 - val_loss: 0.1938 - val_accuracy: 0.9547 - val_precision_5: 0.9700 - val_recall_5: 0.9390\n",
            "Epoch 12/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.1713 - accuracy: 0.9622 - precision_5: 0.9746 - recall_5: 0.9476 - val_loss: 0.1786 - val_accuracy: 0.9578 - val_precision_5: 0.9719 - val_recall_5: 0.9448\n",
            "Epoch 13/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.1582 - accuracy: 0.9642 - precision_5: 0.9757 - recall_5: 0.9520 - val_loss: 0.1676 - val_accuracy: 0.9608 - val_precision_5: 0.9732 - val_recall_5: 0.9488\n",
            "Epoch 14/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.1474 - accuracy: 0.9665 - precision_5: 0.9767 - recall_5: 0.9557 - val_loss: 0.1577 - val_accuracy: 0.9623 - val_precision_5: 0.9726 - val_recall_5: 0.9522\n",
            "Epoch 15/15\n",
            "422/422 [==============================] - 2s 4ms/step - loss: 0.1385 - accuracy: 0.9685 - precision_5: 0.9776 - recall_5: 0.9591 - val_loss: 0.1497 - val_accuracy: 0.9623 - val_precision_5: 0.9726 - val_recall_5: 0.9535\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KlYEJXbQbX3e",
        "outputId": "8343c697-d1de-434a-cf02-8bda4773f054"
      },
      "source": [
        "print(history.history.keys())"
      ],
      "execution_count": 68,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "dict_keys(['loss', 'accuracy', 'precision_5', 'recall_5', 'val_loss', 'val_accuracy', 'val_precision_5', 'val_recall_5'])\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "3w9nUSy1bX3f",
        "outputId": "601283ad-6b8f-4ee5-dabd-36e4f547862b"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title(\"model accuracy\")\n",
        "plt.ylabel(\"accuracy\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.legend(['train','test'],loc='lower right')\n",
        "plt.show()"
      ],
      "execution_count": 69,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "OGQf4_sAbX3g",
        "outputId": "f32b634b-0b81-4d71-9ea6-cf838bb3794e"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title(\"model loss\")\n",
        "plt.ylabel(\"loss\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.legend(['train','test'],loc='upper right')\n",
        "plt.show()"
      ],
      "execution_count": 70,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "btCYNwo2qFeL",
        "outputId": "e38cc310-d6e6-444d-9ce6-5bbf473fd51d"
      },
      "source": [
        "score = model.evaluate(testX, testY, verbose=0)\n",
        "print(\"Test loss:\", score[0])\n",
        "print(\"Test accuracy:\", score[1])"
      ],
      "execution_count": 71,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Test loss: 0.14050395786762238\n",
            "Test accuracy: 0.9699575304985046\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OKYf_OeXGwn_",
        "outputId": "934098df-6741-44e4-bc3f-68b93059ff72"
      },
      "source": [
        "loss, accuracy, precision, recall = model.evaluate(testX, testY, verbose=0)\n",
        "print('Loss', loss)\n",
        "print('Accuracy', accuracy)\n",
        "print('Precision', precision)\n",
        "print('Recall', recall)\n",
        "\n",
        "f1_score = (2* recall*precision)/(recall+precision)\n",
        "print(\"f1_score\", f1_score)"
      ],
      "execution_count": 72,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Loss 0.14050395786762238\n",
            "Accuracy 0.9699575304985046\n",
            "Precision 0.9767484664916992\n",
            "Recall 0.9617575407028198\n",
            "f1_score 0.969195039408556\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vms4pcBpRAr4"
      },
      "source": [
        "## ConvNet"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "y3eYgDEiPWa_",
        "outputId": "5432487b-3587-41f7-c97e-c2ec352fa2fa"
      },
      "source": [
        "model = keras.Sequential(\n",
        "    [\n",
        "        keras.Input(shape=input_shape),\n",
        "        layers.Conv2D(32, kernel_size=(3, 3), activation=\"relu\"),\n",
        "        layers.MaxPooling2D(pool_size=(2, 2)),\n",
        "        layers.Conv2D(64, kernel_size=(3, 3), activation=\"relu\"),\n",
        "        layers.MaxPooling2D(pool_size=(2, 2)),\n",
        "        layers.Flatten(),\n",
        "        layers.Dropout(0.5),\n",
        "        layers.Dense(num_classes, activation=\"softmax\"),\n",
        "    ]\n",
        ")\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_4\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_2 (Conv2D)            (None, 26, 26, 32)        320       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 13, 13, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 11, 11, 64)        18496     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2 (None, 5, 5, 64)          0         \n",
            "_________________________________________________________________\n",
            "flatten_4 (Flatten)          (None, 1600)              0         \n",
            "_________________________________________________________________\n",
            "dropout_3 (Dropout)          (None, 1600)              0         \n",
            "_________________________________________________________________\n",
            "dense_7 (Dense)              (None, 10)                16010     \n",
            "=================================================================\n",
            "Total params: 34,826\n",
            "Trainable params: 34,826\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "u4prc3LmRP3H",
        "outputId": "44c93b74-ddb0-425a-9cb8-ac04019f7fb7"
      },
      "source": [
        "# Scale images to the [0, 1] range\n",
        "trainX = np.array(trainX)\n",
        "testX = np.array(testX)\n",
        "trainX = trainX.astype(\"float32\") / 255\n",
        "testX = testX.astype(\"float32\") / 255\n",
        "\n",
        "# Make sure images have shape (28, 28, 1)\n",
        "trainX = np.expand_dims(trainX, -1)\n",
        "testX = np.expand_dims(testX, -1)\n",
        "\n",
        "print(\"x_train shape:\", trainX.shape)\n",
        "print(trainX.shape[0], \"train samples\")\n",
        "print(testX.shape[0], \"test samples\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "x_train shape: (60000, 28, 28, 1)\n",
            "60000 train samples\n",
            "27561 test samples\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8HNDp2JgRaLI"
      },
      "source": [
        "trainY = np.array(trainY)\n",
        "testY = np.array(testY)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4D57rMIPRaLJ",
        "outputId": "20b8aa11-8111-486a-ef14-930101753d16"
      },
      "source": [
        "batch_size = 128\n",
        "epochs = 15\n",
        "\n",
        "model.compile(loss=\"categorical_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\",keras.metrics.Precision(),keras.metrics.Recall()])\n",
        "\n",
        "history = model.fit(trainX, trainY, batch_size=batch_size, epochs=epochs, validation_split=0.1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/15\n",
            "422/422 [==============================] - 32s 73ms/step - loss: 1.7154 - accuracy: 0.4957 - precision_3: 0.9573 - recall_3: 0.1507 - val_loss: 0.4901 - val_accuracy: 0.9143 - val_precision_3: 0.9653 - val_recall_3: 0.7933\n",
            "Epoch 2/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.3908 - accuracy: 0.8946 - precision_3: 0.9340 - recall_3: 0.8420 - val_loss: 0.2369 - val_accuracy: 0.9388 - val_precision_3: 0.9588 - val_recall_3: 0.9183\n",
            "Epoch 3/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.2771 - accuracy: 0.9211 - precision_3: 0.9394 - recall_3: 0.9010 - val_loss: 0.1847 - val_accuracy: 0.9502 - val_precision_3: 0.9639 - val_recall_3: 0.9388\n",
            "Epoch 4/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.2342 - accuracy: 0.9317 - precision_3: 0.9470 - recall_3: 0.9184 - val_loss: 0.1587 - val_accuracy: 0.9563 - val_precision_3: 0.9683 - val_recall_3: 0.9470\n",
            "Epoch 5/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.2109 - accuracy: 0.9395 - precision_3: 0.9513 - recall_3: 0.9273 - val_loss: 0.1440 - val_accuracy: 0.9590 - val_precision_3: 0.9697 - val_recall_3: 0.9510\n",
            "Epoch 6/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.1938 - accuracy: 0.9453 - precision_3: 0.9554 - recall_3: 0.9353 - val_loss: 0.1304 - val_accuracy: 0.9627 - val_precision_3: 0.9738 - val_recall_3: 0.9558\n",
            "Epoch 7/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.1817 - accuracy: 0.9476 - precision_3: 0.9574 - recall_3: 0.9386 - val_loss: 0.1255 - val_accuracy: 0.9652 - val_precision_3: 0.9736 - val_recall_3: 0.9583\n",
            "Epoch 8/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.1684 - accuracy: 0.9520 - precision_3: 0.9612 - recall_3: 0.9435 - val_loss: 0.1093 - val_accuracy: 0.9715 - val_precision_3: 0.9779 - val_recall_3: 0.9647\n",
            "Epoch 9/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.1539 - accuracy: 0.9566 - precision_3: 0.9645 - recall_3: 0.9494 - val_loss: 0.1013 - val_accuracy: 0.9743 - val_precision_3: 0.9802 - val_recall_3: 0.9668\n",
            "Epoch 10/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.1441 - accuracy: 0.9586 - precision_3: 0.9660 - recall_3: 0.9520 - val_loss: 0.0930 - val_accuracy: 0.9747 - val_precision_3: 0.9805 - val_recall_3: 0.9700\n",
            "Epoch 11/15\n",
            "422/422 [==============================] - 30s 71ms/step - loss: 0.1325 - accuracy: 0.9628 - precision_3: 0.9694 - recall_3: 0.9571 - val_loss: 0.0849 - val_accuracy: 0.9790 - val_precision_3: 0.9835 - val_recall_3: 0.9737\n",
            "Epoch 12/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.1282 - accuracy: 0.9642 - precision_3: 0.9696 - recall_3: 0.9584 - val_loss: 0.0774 - val_accuracy: 0.9813 - val_precision_3: 0.9845 - val_recall_3: 0.9768\n",
            "Epoch 13/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.1208 - accuracy: 0.9662 - precision_3: 0.9718 - recall_3: 0.9613 - val_loss: 0.0727 - val_accuracy: 0.9810 - val_precision_3: 0.9842 - val_recall_3: 0.9775\n",
            "Epoch 14/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.1181 - accuracy: 0.9660 - precision_3: 0.9712 - recall_3: 0.9613 - val_loss: 0.0678 - val_accuracy: 0.9825 - val_precision_3: 0.9852 - val_recall_3: 0.9792\n",
            "Epoch 15/15\n",
            "422/422 [==============================] - 30s 72ms/step - loss: 0.1110 - accuracy: 0.9688 - precision_3: 0.9736 - recall_3: 0.9649 - val_loss: 0.0637 - val_accuracy: 0.9838 - val_precision_3: 0.9859 - val_recall_3: 0.9810\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "S6QFUfrxRaLK",
        "outputId": "167e0f2c-72ff-41ca-fb52-68c62394c97c"
      },
      "source": [
        "print(history.history.keys())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "dict_keys(['loss', 'accuracy', 'precision_3', 'recall_3', 'val_loss', 'val_accuracy', 'val_precision_3', 'val_recall_3'])\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "zdX_PBdXRaLL",
        "outputId": "837e5256-03c1-4f98-b264-e5cbfa121bf3"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title(\"model accuracy\")\n",
        "plt.ylabel(\"accuracy\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.legend(['train','test'],loc='lower right')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "6h8ViHgQRaLL",
        "outputId": "ace93e2d-b4e9-4384-ff2f-4af1f4e489bd"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title(\"model loss\")\n",
        "plt.ylabel(\"loss\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.legend(['train','test'],loc='upper right')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rVDrH9X7RaLM",
        "outputId": "bb126727-6844-4c2a-80e7-00f536c12cde"
      },
      "source": [
        "score = model.evaluate(testX, testY, verbose=0)\n",
        "print(\"Test loss:\", score[0])\n",
        "print(\"Test accuracy:\", score[1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Test loss: 0.0745575949549675\n",
            "Test accuracy: 0.9818947315216064\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ursEkUmZRaLN",
        "outputId": "312ddeb1-76ab-40d3-9907-a540e2d5e4f9"
      },
      "source": [
        "loss, accuracy, precision, recall = model.evaluate(testX, testY, verbose=0)\n",
        "print('Loss', loss)\n",
        "print('Accuracy', accuracy)\n",
        "print('Precision', precision)\n",
        "print('Recall', recall)\n",
        "\n",
        "f1_score = (2* recall*precision)/(recall+precision)\n",
        "print(\"f1_score\", f1_score)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Loss 0.0745575949549675\n",
            "Accuracy 0.9818947315216064\n",
            "Precision 0.9839749336242676\n",
            "Recall 0.9802619814872742\n",
            "f1_score 0.9821149483014988\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6s4xqJzWRaLN"
      },
      "source": [
        "from keras.models import load_model\n",
        "\n",
        "model.save('cnn.h5')  # creates a HDF5 file 'my_model.h5'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "19z_bWXCTWzG"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}