Transfer Learning Cat vs Dogs.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "Course 2 - Part 6 - Lesson 3 - Notebook.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "rX8mhOLljYeM"
      },
      "source": [
        "##### Copyright 2019 The TensorFlow Authors."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "cellView": "form",
        "colab_type": "code",
        "id": "BZSlp3DAjdYf",
        "colab": {}
      },
      "source": [
        "#@title Licensed under the Apache License, Version 2.0 (the \"License\");\n",
        "# you may not use this file except in compliance with the License.\n",
        "# You may obtain a copy of the License at\n",
        "#\n",
        "# https://www.apache.org/licenses/LICENSE-2.0\n",
        "#\n",
        "# Unless required by applicable law or agreed to in writing, software\n",
        "# distributed under the License is distributed on an \"AS IS\" BASIS,\n",
        "# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n",
        "# See the License for the specific language governing permissions and\n",
        "# limitations under the License."
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "1xJZ5glPPCRz",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 241
        },
        "outputId": "1f7fb574-4b10-4662-8631-9ae07d1edf7d"
      },
      "source": [
        "import os\n",
        "\n",
        "from tensorflow.keras import layers\n",
        "from tensorflow.keras import Model\n",
        "!wget --no-check-certificate \\\n",
        "    https://storage.googleapis.com/mledu-datasets/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5 \\\n",
        "    -O /tmp/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
        "  \n",
        "from tensorflow.keras.applications.inception_v3 import InceptionV3\n",
        "\n",
        "local_weights_file = '/tmp/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5'\n",
        "\n",
        "pre_trained_model = InceptionV3(input_shape = (150, 150, 3), \n",
        "                                include_top = False, \n",
        "                                weights = None)\n",
        "\n",
        "pre_trained_model.load_weights(local_weights_file)\n",
        "\n",
        "for layer in pre_trained_model.layers:\n",
        "  layer.trainable = False\n",
        "  \n",
        "# pre_trained_model.summary()\n",
        "\n",
        "last_layer = pre_trained_model.get_layer('mixed7')\n",
        "print('last layer output shape: ', last_layer.output_shape)\n",
        "last_output = last_layer.output"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2020-06-20 13:57:52--  https://storage.googleapis.com/mledu-datasets/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 173.194.76.128, 2a00:1450:400c:c08::80\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|173.194.76.128|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 87910968 (84M) [application/x-hdf]\n",
            "Saving to: ‘/tmp/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5’\n",
            "\n",
            "\r          /tmp/ince   0%[                    ]       0  --.-KB/s               \r         /tmp/incep   9%[>                   ]   8.01M  19.1MB/s               \r        /tmp/incept  57%[==========>         ]  48.01M  72.4MB/s               \r       /tmp/incepti  85%[================>   ]  72.01M  80.5MB/s               \r/tmp/inception_v3_w 100%[===================>]  83.84M  88.9MB/s    in 0.9s    \n",
            "\n",
            "2020-06-20 13:57:53 (88.9 MB/s) - ‘/tmp/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5’ saved [87910968/87910968]\n",
            "\n",
            "last layer output shape:  (None, 7, 7, 768)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "BMXb913pbvFg",
        "colab": {}
      },
      "source": [
        "from tensorflow.keras.optimizers import RMSprop\n",
        "\n",
        "# Flatten the output layer to 1 dimension\n",
        "x = layers.Flatten()(last_output)\n",
        "# Add a fully connected layer with 1,024 hidden units and ReLU activation\n",
        "x = layers.Dense(1024, activation='relu')(x)\n",
        "# Add a dropout rate of 0.2\n",
        "x = layers.Dropout(0.2)(x)                  \n",
        "# Add a final sigmoid layer for classification\n",
        "x = layers.Dense  (1, activation='sigmoid')(x)           \n",
        "\n",
        "model = Model( pre_trained_model.input, x) \n",
        "\n",
        "model.compile(optimizer = RMSprop(lr=0.0001), \n",
        "              loss = 'binary_crossentropy', \n",
        "              metrics = ['accuracy'])\n"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "O4s8HckqGlnb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 238
        },
        "outputId": "9cfc75fa-cd60-40ad-d896-666201e1ac94"
      },
      "source": [
        "!wget --no-check-certificate \\\n",
        "        https://storage.googleapis.com/mledu-datasets/cats_and_dogs_filtered.zip \\\n",
        "       -O /tmp/cats_and_dogs_filtered.zip\n",
        "\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "\n",
        "import os\n",
        "import zipfile\n",
        "\n",
        "local_zip = '//tmp/cats_and_dogs_filtered.zip'\n",
        "\n",
        "zip_ref = zipfile.ZipFile(local_zip, 'r')\n",
        "\n",
        "zip_ref.extractall('/tmp')\n",
        "zip_ref.close()\n",
        "\n",
        "# Define our example directories and files\n",
        "base_dir = '/tmp/cats_and_dogs_filtered'\n",
        "\n",
        "train_dir = os.path.join( base_dir, 'train')\n",
        "validation_dir = os.path.join( base_dir, 'validation')\n",
        "\n",
        "\n",
        "train_cats_dir = os.path.join(train_dir, 'cats') # Directory with our training cat pictures\n",
        "train_dogs_dir = os.path.join(train_dir, 'dogs') # Directory with our training dog pictures\n",
        "validation_cats_dir = os.path.join(validation_dir, 'cats') # Directory with our validation cat pictures\n",
        "validation_dogs_dir = os.path.join(validation_dir, 'dogs')# Directory with our validation dog pictures\n",
        "\n",
        "train_cat_fnames = os.listdir(train_cats_dir)\n",
        "train_dog_fnames = os.listdir(train_dogs_dir)\n",
        "\n",
        "# Add our data-augmentation parameters to ImageDataGenerator\n",
        "train_datagen = ImageDataGenerator(rescale = 1./255.,\n",
        "                                   rotation_range = 40,\n",
        "                                   width_shift_range = 0.2,\n",
        "                                   height_shift_range = 0.2,\n",
        "                                   shear_range = 0.2,\n",
        "                                   zoom_range = 0.2,\n",
        "                                   horizontal_flip = True)\n",
        "\n",
        "# Note that the validation data should not be augmented!\n",
        "test_datagen = ImageDataGenerator( rescale = 1.0/255. )\n",
        "\n",
        "# Flow training images in batches of 20 using train_datagen generator\n",
        "train_generator = train_datagen.flow_from_directory(train_dir,\n",
        "                                                    batch_size = 20,\n",
        "                                                    class_mode = 'binary', \n",
        "                                                    target_size = (150, 150))     \n",
        "\n",
        "# Flow validation images in batches of 20 using test_datagen generator\n",
        "validation_generator =  test_datagen.flow_from_directory( validation_dir,\n",
        "                                                          batch_size  = 20,\n",
        "                                                          class_mode  = 'binary', \n",
        "                                                          target_size = (150, 150))"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2020-06-20 13:59:00--  https://storage.googleapis.com/mledu-datasets/cats_and_dogs_filtered.zip\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 173.194.76.128, 2a00:1450:400c:c09::80\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|173.194.76.128|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 68606236 (65M) [application/zip]\n",
            "Saving to: ‘/tmp/cats_and_dogs_filtered.zip’\n",
            "\n",
            "/tmp/cats_and_dogs_ 100%[===================>]  65.43M  82.1MB/s    in 0.8s    \n",
            "\n",
            "2020-06-20 13:59:01 (82.1 MB/s) - ‘/tmp/cats_and_dogs_filtered.zip’ saved [68606236/68606236]\n",
            "\n",
            "Found 2000 images belonging to 2 classes.\n",
            "Found 1000 images belonging to 2 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "Blhq2MAUeyGA",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 697
        },
        "outputId": "72137d40-caac-4bb2-ca9a-a5dcfcfb06ad"
      },
      "source": [
        "history = model.fit(\n",
        "            train_generator,\n",
        "            validation_data = validation_generator,\n",
        "            steps_per_epoch = 100,\n",
        "            epochs = 20,\n",
        "            validation_steps = 50,\n",
        "            verbose = 2)"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "100/100 - 22s - loss: 0.3493 - accuracy: 0.8695 - val_loss: 0.1061 - val_accuracy: 0.9580\n",
            "Epoch 2/20\n",
            "100/100 - 21s - loss: 0.2183 - accuracy: 0.9155 - val_loss: 0.1149 - val_accuracy: 0.9580\n",
            "Epoch 3/20\n",
            "100/100 - 21s - loss: 0.1954 - accuracy: 0.9275 - val_loss: 0.1309 - val_accuracy: 0.9570\n",
            "Epoch 4/20\n",
            "100/100 - 21s - loss: 0.1901 - accuracy: 0.9270 - val_loss: 0.1480 - val_accuracy: 0.9540\n",
            "Epoch 5/20\n",
            "100/100 - 21s - loss: 0.1819 - accuracy: 0.9390 - val_loss: 0.2095 - val_accuracy: 0.9360\n",
            "Epoch 6/20\n",
            "100/100 - 21s - loss: 0.2199 - accuracy: 0.9240 - val_loss: 0.0968 - val_accuracy: 0.9680\n",
            "Epoch 7/20\n",
            "100/100 - 21s - loss: 0.1691 - accuracy: 0.9490 - val_loss: 0.1512 - val_accuracy: 0.9570\n",
            "Epoch 8/20\n",
            "100/100 - 22s - loss: 0.1564 - accuracy: 0.9470 - val_loss: 0.1405 - val_accuracy: 0.9580\n",
            "Epoch 9/20\n",
            "100/100 - 21s - loss: 0.1589 - accuracy: 0.9440 - val_loss: 0.1403 - val_accuracy: 0.9570\n",
            "Epoch 10/20\n",
            "100/100 - 21s - loss: 0.1657 - accuracy: 0.9465 - val_loss: 0.1141 - val_accuracy: 0.9660\n",
            "Epoch 11/20\n",
            "100/100 - 21s - loss: 0.1571 - accuracy: 0.9480 - val_loss: 0.3424 - val_accuracy: 0.9190\n",
            "Epoch 12/20\n",
            "100/100 - 21s - loss: 0.1400 - accuracy: 0.9530 - val_loss: 0.1335 - val_accuracy: 0.9640\n",
            "Epoch 13/20\n",
            "100/100 - 21s - loss: 0.1469 - accuracy: 0.9490 - val_loss: 0.1379 - val_accuracy: 0.9670\n",
            "Epoch 14/20\n",
            "100/100 - 21s - loss: 0.1722 - accuracy: 0.9425 - val_loss: 0.2657 - val_accuracy: 0.9350\n",
            "Epoch 15/20\n",
            "100/100 - 22s - loss: 0.1411 - accuracy: 0.9505 - val_loss: 0.1348 - val_accuracy: 0.9640\n",
            "Epoch 16/20\n",
            "100/100 - 21s - loss: 0.1401 - accuracy: 0.9545 - val_loss: 0.1537 - val_accuracy: 0.9660\n",
            "Epoch 17/20\n",
            "100/100 - 21s - loss: 0.1387 - accuracy: 0.9455 - val_loss: 0.1129 - val_accuracy: 0.9670\n",
            "Epoch 18/20\n",
            "100/100 - 21s - loss: 0.1591 - accuracy: 0.9500 - val_loss: 0.1141 - val_accuracy: 0.9630\n",
            "Epoch 19/20\n",
            "100/100 - 21s - loss: 0.1268 - accuracy: 0.9570 - val_loss: 0.1238 - val_accuracy: 0.9650\n",
            "Epoch 20/20\n",
            "100/100 - 21s - loss: 0.1415 - accuracy: 0.9610 - val_loss: 0.1178 - val_accuracy: 0.9620\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "C2Fp6Se9rKuL",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 298
        },
        "outputId": "72d9df2a-d05a-4d4d-b5b1-fbb3f32d9ac8"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(len(acc))\n",
        "\n",
        "plt.plot(epochs, acc, 'r', label='Training accuracy')\n",
        "plt.plot(epochs, val_acc, 'b', label='Validation accuracy')\n",
        "plt.title('Training and validation accuracy')\n",
        "plt.legend(loc=0)\n",
        "plt.figure()\n",
        "\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "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"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 0 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}