tf_cats&dogs.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "transfer_learning cats&dogs reloaded.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "VoENcMqH_C1T"
      },
      "source": [
        "# Based on https://www.tensorflow.org/tutorials/images/transfer_learning"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TqOt6Sv7AsMi"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import os\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras.preprocessing import image_dataset_from_directory"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SYxCkyCj1ffb"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')\n",
        "# Mount Drive\n",
        "!unzip /content/drive/MyDrive/TransferLearnImages.zip\n",
        "\n",
        "!mv ./TransferLearnImages ./train\n",
        "!mkdir -p ./validation/Football\n",
        "!mkdir -p ./validation/Racket\n",
        "!mv ./train/Football/*1* ./validation/Football/\n",
        "!mv ./train/Racket/*1* ./validation/Racket/"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "B1eh0hx4w5zQ",
        "outputId": "f9ce1e98-3c66-4f78-8814-870ce3f31741"
      },
      "source": [
        "PATH = \"/content/\"\r\n",
        "\r\n",
        "train_dir = os.path.join(PATH, 'train')\r\n",
        "validation_dir = os.path.join(PATH, 'validation')\r\n",
        "\r\n",
        "BATCH_SIZE = 32\r\n",
        "IMG_SIZE = (160, 160)\r\n",
        "\r\n",
        "train_dataset = image_dataset_from_directory(train_dir,\r\n",
        "                                             shuffle=True,\r\n",
        "                                             batch_size=BATCH_SIZE,\r\n",
        "                                             image_size=IMG_SIZE)\r\n",
        "\r\n",
        "validation_dataset = image_dataset_from_directory(validation_dir,\r\n",
        "                                                  shuffle=True,\r\n",
        "                                                  batch_size=BATCH_SIZE,\r\n",
        "                                                  image_size=IMG_SIZE)\r\n",
        "class_names = train_dataset.class_names"
      ],
      "execution_count": 88,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 149 files belonging to 2 classes.\n",
            "Found 39 files belonging to 2 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0Ddq4kFcyMh_"
      },
      "source": [
        "AUTOTUNE = tf.data.experimental.AUTOTUNE\r\n",
        "train_dataset = train_dataset.prefetch(buffer_size=AUTOTUNE)\r\n",
        "validation_dataset = validation_dataset.prefetch(buffer_size=AUTOTUNE)"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3P99QiMGit1A"
      },
      "source": [
        "data_augmentation = tf.keras.Sequential([\n",
        "  tf.keras.layers.experimental.preprocessing.RandomFlip('horizontal'),\n",
        "  tf.keras.layers.experimental.preprocessing.RandomRotation(0.2),\n",
        "])"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cO0HM9JAQUFq"
      },
      "source": [
        "preprocess_input = tf.keras.applications.mobilenet_v2.preprocess_input\r\n",
        "\r\n",
        "rescale = tf.keras.layers.experimental.preprocessing.Rescaling(1./127.5, offset= -1)"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "19IQ2gqneqmS",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e2b79bec-7035-4a2c-fef7-f7ef31016ba7"
      },
      "source": [
        "# Create the base model from the pre-trained model MobileNet V2\n",
        "IMG_SHAPE = IMG_SIZE + (3,)\n",
        "base_model = tf.keras.applications.MobileNetV2(input_shape=IMG_SHAPE,\n",
        "                                               include_top=False,\n",
        "                                               weights='imagenet')"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/mobilenet_v2/mobilenet_v2_weights_tf_dim_ordering_tf_kernels_1.0_160_no_top.h5\n",
            "9412608/9406464 [==============================] - 0s 0us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Y-2LJL0EEUcx",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "80692947-8703-4bf8-e939-f11bd9868288"
      },
      "source": [
        "image_batch, label_batch = next(iter(train_dataset))\n",
        "feature_batch = base_model(image_batch)\n",
        "print(feature_batch.shape)"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(32, 5, 5, 1280)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OTCJH4bphOeo"
      },
      "source": [
        "base_model.trainable = False"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dLnpMF5KOALm",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "61587145-be3e-411d-be82-af4daba428f5"
      },
      "source": [
        "global_average_layer = tf.keras.layers.GlobalAveragePooling2D()\n",
        "feature_batch_average = global_average_layer(feature_batch)\n",
        "print(feature_batch_average.shape)"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(32, 1280)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Wv4afXKj6cVa",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "93d4daa9-d754-4624-917e-8e8d918028b1"
      },
      "source": [
        "prediction_layer = tf.keras.layers.Dense(1)\n",
        "prediction_batch = prediction_layer(feature_batch_average)\n",
        "print(prediction_batch.shape)"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(32, 1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DgzQX6Veb2WT"
      },
      "source": [
        "inputs = tf.keras.Input(shape=(160, 160, 3))\n",
        "x = data_augmentation(inputs)\n",
        "x = preprocess_input(x)\n",
        "x = base_model(x, training=False)\n",
        "x = global_average_layer(x)\n",
        "x = tf.keras.layers.Dropout(0.2)(x)\n",
        "outputs = prediction_layer(x)\n",
        "model = tf.keras.Model(inputs, outputs)"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RpR8HdyMhukJ"
      },
      "source": [
        "base_learning_rate = 0.0001\n",
        "model.compile(optimizer=tf.keras.optimizers.Adam(lr=base_learning_rate),\n",
        "              loss=tf.keras.losses.BinaryCrossentropy(from_logits=True),\n",
        "              metrics=['accuracy'])"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Om4O3EESkab1",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ce85fe9e-9d79-4ea4-d90f-fb0639172b39"
      },
      "source": [
        "initial_epochs = 10\r\n",
        "loss0, accuracy0 = model.evaluate(validation_dataset)"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2/2 [==============================] - 0s 103ms/step - loss: 0.6086 - accuracy: 0.5641\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JsaRFlZ9B6WK",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "02e1acf1-f758-4b48-fcba-9b2c9a4b0cd1"
      },
      "source": [
        "history = model.fit(train_dataset,\n",
        "                    epochs=initial_epochs,\n",
        "                    validation_data=validation_dataset)"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "5/5 [==============================] - 0s 49ms/step - loss: 0.6504 - accuracy: 0.6376 - val_loss: 0.5804 - val_accuracy: 0.5897\n",
            "Epoch 2/10\n",
            "5/5 [==============================] - 0s 46ms/step - loss: 0.6183 - accuracy: 0.7047 - val_loss: 0.5545 - val_accuracy: 0.6410\n",
            "Epoch 3/10\n",
            "5/5 [==============================] - 0s 42ms/step - loss: 0.5501 - accuracy: 0.7584 - val_loss: 0.5307 - val_accuracy: 0.6667\n",
            "Epoch 4/10\n",
            "5/5 [==============================] - 0s 42ms/step - loss: 0.5525 - accuracy: 0.7114 - val_loss: 0.5088 - val_accuracy: 0.7436\n",
            "Epoch 5/10\n",
            "5/5 [==============================] - 0s 44ms/step - loss: 0.5314 - accuracy: 0.7584 - val_loss: 0.4883 - val_accuracy: 0.7436\n",
            "Epoch 6/10\n",
            "5/5 [==============================] - 0s 44ms/step - loss: 0.5358 - accuracy: 0.7248 - val_loss: 0.4679 - val_accuracy: 0.7436\n",
            "Epoch 7/10\n",
            "5/5 [==============================] - 0s 39ms/step - loss: 0.4824 - accuracy: 0.7315 - val_loss: 0.4484 - val_accuracy: 0.7949\n",
            "Epoch 8/10\n",
            "5/5 [==============================] - 0s 45ms/step - loss: 0.4788 - accuracy: 0.7852 - val_loss: 0.4293 - val_accuracy: 0.7949\n",
            "Epoch 9/10\n",
            "5/5 [==============================] - 0s 45ms/step - loss: 0.4361 - accuracy: 0.8255 - val_loss: 0.4115 - val_accuracy: 0.8205\n",
            "Epoch 10/10\n",
            "5/5 [==============================] - 0s 41ms/step - loss: 0.4335 - accuracy: 0.8523 - val_loss: 0.3951 - val_accuracy: 0.8205\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4nzcagVitLQm"
      },
      "source": [
        "base_model.trainable = True"
      ],
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-4HgVAacRs5v",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "fd0e2fde-a46c-475b-e662-64d8dff153a5"
      },
      "source": [
        "# Let's take a look to see how many layers are in the base model\n",
        "print(\"Number of layers in the base model: \", len(base_model.layers))\n",
        "\n",
        "# Fine-tune from this layer onwards\n",
        "fine_tune_at = 100\n",
        "\n",
        "# Freeze all the layers before the `fine_tune_at` layer\n",
        "for layer in base_model.layers[:fine_tune_at]:\n",
        "  layer.trainable =  False"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Number of layers in the base model:  155\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NtUnaz0WUDva"
      },
      "source": [
        "model.compile(loss=tf.keras.losses.BinaryCrossentropy(from_logits=True),\n",
        "              optimizer = tf.keras.optimizers.RMSprop(lr=base_learning_rate/10),\n",
        "              metrics=['accuracy'])"
      ],
      "execution_count": 25,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ECQLkAsFTlun",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1948e765-2599-447b-dabf-04d8b366e62d"
      },
      "source": [
        "fine_tune_epochs = 10\n",
        "total_epochs =  initial_epochs + fine_tune_epochs\n",
        "\n",
        "history_fine = model.fit(train_dataset,\n",
        "                         epochs=total_epochs,\n",
        "                         initial_epoch=history.epoch[-1],\n",
        "                         validation_data=validation_dataset)"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 10/20\n",
            "5/5 [==============================] - 1s 183ms/step - loss: 0.0037 - accuracy: 1.0000 - val_loss: 0.0174 - val_accuracy: 1.0000\n",
            "Epoch 11/20\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 0.0021 - accuracy: 1.0000 - val_loss: 0.0157 - val_accuracy: 1.0000\n",
            "Epoch 12/20\n",
            "5/5 [==============================] - 0s 50ms/step - loss: 0.0017 - accuracy: 1.0000 - val_loss: 0.0146 - val_accuracy: 1.0000\n",
            "Epoch 13/20\n",
            "5/5 [==============================] - 0s 51ms/step - loss: 9.8723e-04 - accuracy: 1.0000 - val_loss: 0.0125 - val_accuracy: 1.0000\n",
            "Epoch 14/20\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0011 - accuracy: 1.0000 - val_loss: 0.0139 - val_accuracy: 1.0000\n",
            "Epoch 15/20\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.0012 - accuracy: 1.0000 - val_loss: 0.0122 - val_accuracy: 1.0000\n",
            "Epoch 16/20\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 3.9817e-04 - accuracy: 1.0000 - val_loss: 0.0117 - val_accuracy: 1.0000\n",
            "Epoch 17/20\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 5.6107e-04 - accuracy: 1.0000 - val_loss: 0.0117 - val_accuracy: 1.0000\n",
            "Epoch 18/20\n",
            "5/5 [==============================] - 0s 51ms/step - loss: 9.4691e-04 - accuracy: 1.0000 - val_loss: 0.0249 - val_accuracy: 1.0000\n",
            "Epoch 19/20\n",
            "5/5 [==============================] - 0s 51ms/step - loss: 5.8004e-04 - accuracy: 1.0000 - val_loss: 0.0156 - val_accuracy: 1.0000\n",
            "Epoch 20/20\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 2.7880e-04 - accuracy: 1.0000 - val_loss: 0.0118 - val_accuracy: 1.0000\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 473
        },
        "id": "hm9eWdjSzXmZ",
        "outputId": "bdae8101-1387-406a-e30b-b02b225635fd"
      },
      "source": [
        "# test model with new image\r\n",
        "!wget -O image.jpg https://image.shutterstock.com/image-photo/soccer-ball-isolated-on-white-260nw-400438939.jpg\r\n",
        "#!wget -O image.jpg https://www.tennisnuts.com/images/product/full/102334-136_babolat-pure-drive_angle.jpg\r\n",
        "\r\n",
        "# process image for model\r\n",
        "\r\n",
        "img = image.load_img('image.jpg', target_size=IMG_SIZE)\r\n",
        "x = image.img_to_array(img)\r\n",
        "x = backend.expand_dims(x, axis=0)\r\n",
        "\r\n",
        "# display image\r\n",
        "import matplotlib.pyplot as plt\r\n",
        "plt.imshow(img)\r\n",
        "\r\n",
        "# run prediction\r\n",
        "prediction = model.predict(x)\r\n",
        "prediction = tf.nn.sigmoid(prediction)\r\n",
        "prediction = tf.where(prediction < 0.5, 0, 1)\r\n",
        "print(class_names[prediction.numpy()[0][0]])"
      ],
      "execution_count": 87,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2020-12-11 22:40:47--  https://image.shutterstock.com/image-photo/soccer-ball-isolated-on-white-260nw-400438939.jpg\n",
            "Resolving image.shutterstock.com (image.shutterstock.com)... 23.36.53.150\n",
            "Connecting to image.shutterstock.com (image.shutterstock.com)|23.36.53.150|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 9309 (9.1K) [image/jpeg]\n",
            "Saving to: ‘image.jpg’\n",
            "\n",
            "\rimage.jpg             0%[                    ]       0  --.-KB/s               \rimage.jpg           100%[===================>]   9.09K  --.-KB/s    in 0s      \n",
            "\n",
            "2020-12-11 22:40:47 (139 MB/s) - ‘image.jpg’ saved [9309/9309]\n",
            "\n",
            "Football\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}