cnn_transfer_learning.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "8L9VMKsIf_vm"
   },
   "source": [
    "Transfer Learning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "oFr4_S7eiMJq"
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import tensorflow_datasets as tfds\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "#plt.gray()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "NkDPHomNa6fo",
    "outputId": "6eb364cd-e238-46b7-f408-1a6e3a1528ab"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg19/vgg19_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
      "80142336/80134624 [==============================] - 0s 0us/step\n",
      "80150528/80134624 [==============================] - 0s 0us/step\n"
     ]
    }
   ],
   "source": [
    "vgg19 = tf.keras.applications.VGG19(\n",
    "    include_top=False,\n",
    "    weights=\"imagenet\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "4DMmwhKvgBpM"
   },
   "outputs": [],
   "source": [
    "#tf.keras.applications.vgg19.preprocess_input"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "5VdbRfrTimuZ",
    "outputId": "9e30158d-4d4a-4af0-ef99-79c4a08ba590"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mounted at /gdrive\n"
     ]
    }
   ],
   "source": [
    "from google.colab import drive\n",
    "drive.mount('/gdrive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "Qm9YPSS4im9q"
   },
   "outputs": [],
   "source": [
    "!mkdir datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "OjF58b2xirdP"
   },
   "outputs": [],
   "source": [
    "!cp /gdrive/MyDrive/101_ObjectCategories.tar.gz ./datasets/101_ObjectCategories.tar.gz"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "EGwEIL2Pisth"
   },
   "outputs": [],
   "source": [
    "!tar -zxvf ./datasets/101_ObjectCategories.tar.gz"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "D9noAvzviuHV"
   },
   "outputs": [],
   "source": [
    "data_dir = './101_ObjectCategories'\n",
    "height = 224\n",
    "width = 224"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "QOzQcQg1iwlU",
    "outputId": "83cf7c94-7472-4d43-95d0-370743b1d096"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 9144 files belonging to 102 classes.\n",
      "Using 7316 files for training.\n",
      "Found 9144 files belonging to 102 classes.\n",
      "Using 1828 files for validation.\n"
     ]
    }
   ],
   "source": [
    "train_dataset = tf.keras.preprocessing.image_dataset_from_directory(\n",
    "    data_dir,\n",
    "    labels=\"inferred\",\n",
    "    label_mode=\"categorical\",\n",
    "    class_names=None,\n",
    "    color_mode=\"rgb\",\n",
    "    batch_size=16,\n",
    "    image_size=(height, width),\n",
    "    shuffle=True,\n",
    "    seed=1,\n",
    "    validation_split=0.2,\n",
    "    subset=\"training\",\n",
    "    interpolation=\"bilinear\",\n",
    "    follow_links=False,\n",
    "    crop_to_aspect_ratio=False\n",
    ").map(lambda x,y: (tf.keras.applications.vgg19.preprocess_input(x), y))\n",
    "\n",
    "val_dataset = tf.keras.preprocessing.image_dataset_from_directory(\n",
    "    data_dir,\n",
    "    labels=\"inferred\",\n",
    "    label_mode=\"categorical\",\n",
    "    class_names=None,\n",
    "    color_mode=\"rgb\",\n",
    "    batch_size=16,\n",
    "    image_size=(height, width),\n",
    "    shuffle=False,\n",
    "    seed=1,\n",
    "    validation_split=0.2,\n",
    "    subset=\"validation\",\n",
    "    interpolation=\"bilinear\",\n",
    "    follow_links=False,\n",
    "    crop_to_aspect_ratio=False\n",
    ").map(lambda x,y: (tf.keras.applications.vgg19.preprocess_input(x), y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 408
    },
    "id": "1hKlluNuix2F",
    "outputId": "6b79e259-8b1c-4203-89e2-a7348d50c8a5"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tf.Tensor(\n",
      "[1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
      " 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
      " 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
      " 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
      " 0. 0. 0. 0. 0. 0.], shape=(102,), dtype=float32)\n",
      "tf.Tensor(0, shape=(), dtype=int64)\n"
     ]
    }
   ],
   "source": [
    "for x,y in train_dataset.take(1):\n",
    "  plt.imshow(x[0])\n",
    "  plt.show()\n",
    "\n",
    "  print(y[0])\n",
    "  print(tf.argmax(y[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "rfzb8e0_izM5"
   },
   "outputs": [],
   "source": [
    "def plot_history(h):\n",
    "  plt.plot(h.history['accuracy'])\n",
    "  plt.plot(h.history['val_accuracy'])\n",
    "  plt.title('model accuracy')\n",
    "  plt.ylabel('accuracy')\n",
    "  plt.xlabel('epoch')\n",
    "  plt.legend(['train', 'val'], loc='upper left')\n",
    "  plt.show()\n",
    "\n",
    "  plt.plot(h.history['loss'])\n",
    "  plt.plot(h.history['val_loss'])\n",
    "  plt.title('model loss')\n",
    "  plt.ylabel('loss')\n",
    "  plt.xlabel('epoch')\n",
    "  plt.legend(['train', 'val'], loc='upper left')\n",
    "  plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "w6j_LQc4kGTy"
   },
   "outputs": [],
   "source": [
    "vgg19.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "ZV5TmFRPkS9O",
    "outputId": "255d7144-4900-44c6-c2c5-51d0868ca234"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "input_1 False\n",
      "block1_conv1 False\n",
      "block1_conv2 False\n",
      "block1_pool False\n",
      "block2_conv1 False\n",
      "block2_conv2 False\n",
      "block2_pool False\n",
      "block3_conv1 False\n",
      "block3_conv2 False\n",
      "block3_conv3 False\n",
      "block3_conv4 False\n",
      "block3_pool False\n",
      "block4_conv1 False\n",
      "block4_conv2 False\n",
      "block4_conv3 False\n",
      "block4_conv4 False\n",
      "block4_pool False\n",
      "block5_conv1 False\n",
      "block5_conv2 False\n",
      "block5_conv3 False\n",
      "block5_conv4 False\n",
      "block5_pool False\n"
     ]
    }
   ],
   "source": [
    "for l in vgg19.layers:\n",
    "  print(l.name, l.trainable)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "W4nak-Makko4",
    "outputId": "b5581b1e-c6b9-46f2-fd88-c4cdfc33deec"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"vgg19\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_1 (InputLayer)        [(None, None, None, 3)]   0         \n",
      "                                                                 \n",
      " block1_conv1 (Conv2D)       (None, None, None, 64)    1792      \n",
      "                                                                 \n",
      " block1_conv2 (Conv2D)       (None, None, None, 64)    36928     \n",
      "                                                                 \n",
      " block1_pool (MaxPooling2D)  (None, None, None, 64)    0         \n",
      "                                                                 \n",
      " block2_conv1 (Conv2D)       (None, None, None, 128)   73856     \n",
      "                                                                 \n",
      " block2_conv2 (Conv2D)       (None, None, None, 128)   147584    \n",
      "                                                                 \n",
      " block2_pool (MaxPooling2D)  (None, None, None, 128)   0         \n",
      "                                                                 \n",
      " block3_conv1 (Conv2D)       (None, None, None, 256)   295168    \n",
      "                                                                 \n",
      " block3_conv2 (Conv2D)       (None, None, None, 256)   590080    \n",
      "                                                                 \n",
      " block3_conv3 (Conv2D)       (None, None, None, 256)   590080    \n",
      "                                                                 \n",
      " block3_conv4 (Conv2D)       (None, None, None, 256)   590080    \n",
      "                                                                 \n",
      " block3_pool (MaxPooling2D)  (None, None, None, 256)   0         \n",
      "                                                                 \n",
      " block4_conv1 (Conv2D)       (None, None, None, 512)   1180160   \n",
      "                                                                 \n",
      " block4_conv2 (Conv2D)       (None, None, None, 512)   2359808   \n",
      "                                                                 \n",
      " block4_conv3 (Conv2D)       (None, None, None, 512)   2359808   \n",
      "                                                                 \n",
      " block4_conv4 (Conv2D)       (None, None, None, 512)   2359808   \n",
      "                                                                 \n",
      " block4_pool (MaxPooling2D)  (None, None, None, 512)   0         \n",
      "                                                                 \n",
      " block5_conv1 (Conv2D)       (None, None, None, 512)   2359808   \n",
      "                                                                 \n",
      " block5_conv2 (Conv2D)       (None, None, None, 512)   2359808   \n",
      "                                                                 \n",
      " block5_conv3 (Conv2D)       (None, None, None, 512)   2359808   \n",
      "                                                                 \n",
      " block5_conv4 (Conv2D)       (None, None, None, 512)   2359808   \n",
      "                                                                 \n",
      " block5_pool (MaxPooling2D)  (None, None, None, 512)   0         \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 20,024,384\n",
      "Trainable params: 0\n",
      "Non-trainable params: 20,024,384\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "vgg19.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "ve9w3vu_kprU",
    "outputId": "ff2d4176-c69d-4faa-d1ee-18806840bfa9"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<KerasTensor: shape=(None, None, None, 512) dtype=float32 (created by layer 'block5_pool')>]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vgg19.outputs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "-A90q1BUi2a1",
    "outputId": "50697abe-98db-4807-9f89-91eb4ed54641"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_2\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_4 (InputLayer)        [(None, 224, 224, 3)]     0         \n",
      "                                                                 \n",
      " vgg19 (Functional)          (None, None, None, 512)   20024384  \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 25088)             0         \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 102)               2559078   \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 22,583,462\n",
      "Trainable params: 2,559,078\n",
      "Non-trainable params: 20,024,384\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "input_im = tf.keras.Input(shape=(height, width, 3))\n",
    "x = vgg19(input_im)\n",
    "x = tf.keras.layers.Flatten()(x)\n",
    "x = tf.keras.layers.Dense(102, activation='softmax', kernel_initializer='he_uniform')(x)\n",
    "\n",
    "good_model_adam = tf.keras.Model(input_im, x)\n",
    "good_model_adam.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 468
    },
    "id": "GULGRO5kkx5O",
    "outputId": "dc1761ee-7010-419a-f5d8-89e3c1b5fc07"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "458/458 [==============================] - 40s 68ms/step - loss: 5.7268 - accuracy: 0.6771 - val_loss: 0.4880 - val_accuracy: 0.9508\n",
      "Epoch 2/20\n",
      "458/458 [==============================] - 35s 75ms/step - loss: 0.1543 - accuracy: 0.9773 - val_loss: 0.3746 - val_accuracy: 0.9710\n",
      "Epoch 3/20\n",
      " 60/458 [==>...........................] - ETA: 21s - loss: 0.0518 - accuracy: 0.9958"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "ignored",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-27-a01a5d8adb35>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     11\u001b[0m     metrics='accuracy')\n\u001b[1;32m     12\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 13\u001b[0;31m \u001b[0mh_good_adam\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mgood_model_adam\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain_dataset\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalidation_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mval_dataset\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m20\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/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     62\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     63\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---> 64\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     65\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;31m# pylint: disable=broad-except\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     66\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/python3.7/dist-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_batch_size, validation_freq, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   1214\u001b[0m                 _r=1):\n\u001b[1;32m   1215\u001b[0m               \u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_train_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-> 1216\u001b[0;31m               \u001b[0mtmp_logs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain_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   1217\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   1218\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/python3.7/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/python3.7/dist-packages/tensorflow/python/eager/def_function.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    908\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    909\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--> 910\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    911\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    912\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/python3.7/dist-packages/tensorflow/python/eager/def_function.py\u001b[0m in \u001b[0;36m_call\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    940\u001b[0m       \u001b[0;31m# In this case we have created variables on the first call, so we run the\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    941\u001b[0m       \u001b[0;31m# defunned version which is guaranteed to never create variables.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 942\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_stateless_fn\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;31m# pylint: disable=not-callable\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    943\u001b[0m     \u001b[0;32melif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_stateful_fn\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\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[1;32m    944\u001b[0m       \u001b[0;31m# Release the lock early so that multiple threads can perform the call\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/python3.7/dist-packages/tensorflow/python/eager/function.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   3129\u001b[0m        filtered_flat_args) = self._maybe_define_function(args, kwargs)\n\u001b[1;32m   3130\u001b[0m     return graph_function._call_flat(\n\u001b[0;32m-> 3131\u001b[0;31m         filtered_flat_args, captured_inputs=graph_function.captured_inputs)  # pylint: disable=protected-access\n\u001b[0m\u001b[1;32m   3132\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3133\u001b[0m   \u001b[0;34m@\u001b[0m\u001b[0mproperty\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\u001b[0m in \u001b[0;36m_call_flat\u001b[0;34m(self, args, captured_inputs, cancellation_manager)\u001b[0m\n\u001b[1;32m   1958\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[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1959\u001b[0m       return self._build_call_outputs(self._inference_function.call(\n\u001b[0;32m-> 1960\u001b[0;31m           ctx, args, cancellation_manager=cancellation_manager))\n\u001b[0m\u001b[1;32m   1961\u001b[0m     forward_backward = self._select_forward_and_backward_functions(\n\u001b[1;32m   1962\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/python3.7/dist-packages/tensorflow/python/eager/function.py\u001b[0m in \u001b[0;36mcall\u001b[0;34m(self, ctx, args, cancellation_manager)\u001b[0m\n\u001b[1;32m    601\u001b[0m               \u001b[0minputs\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[1;32m    602\u001b[0m               \u001b[0mattrs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mattrs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 603\u001b[0;31m               ctx=ctx)\n\u001b[0m\u001b[1;32m    604\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    605\u001b[0m           outputs = execute.execute_with_cancellation(\n",
      "\u001b[0;32m/usr/local/lib/python3.7/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     57\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[1;32m     58\u001b[0m     tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,\n\u001b[0;32m---> 59\u001b[0;31m                                         inputs, attrs, num_outputs)\n\u001b[0m\u001b[1;32m     60\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[1;32m     61\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\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;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "optimizer = tf.keras.optimizers.Adam(\n",
    "    learning_rate=0.0001,\n",
    "    beta_1=0.9,\n",
    "    beta_2=0.999,\n",
    "    epsilon=1e-07\n",
    ")\n",
    "\n",
    "good_model_adam.compile(\n",
    "    optimizer=optimizer,\n",
    "    loss='categorical_crossentropy',\n",
    "    metrics='accuracy')\n",
    "\n",
    "h_good_adam = good_model_adam.fit(train_dataset, validation_data = val_dataset, epochs = 20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "V1bO_KdYleu6",
    "outputId": "527ee88a-0fdc-4ab9-a90c-7368537d7e1b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_3\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_5 (InputLayer)        [(None, 224, 224, 3)]     0         \n",
      "                                                                 \n",
      " vgg19 (Functional)          (None, None, None, 512)   20024384  \n",
      "                                                                 \n",
      " global_max_pooling2d (Globa  (None, 512)              0         \n",
      " lMaxPooling2D)                                                  \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 102)               52326     \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 20,076,710\n",
      "Trainable params: 52,326\n",
      "Non-trainable params: 20,024,384\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "input_im = tf.keras.Input(shape=(height, width, 3))\n",
    "x = vgg19(input_im)\n",
    "x = tf.keras.layers.GlobalMaxPooling2D()(x)\n",
    "x = tf.keras.layers.Dense(102, activation='softmax', kernel_initializer='he_uniform')(x)\n",
    "\n",
    "good_model_2_adam = tf.keras.Model(input_im, x)\n",
    "good_model_2_adam.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "background_save": true,
     "base_uri": "https://localhost:8080/"
    },
    "id": "qJSSXG7Kmc6o",
    "outputId": "3df3154a-4f7f-4109-9111-83bc56583f0d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "458/458 [==============================] - 31s 66ms/step - loss: 64.5917 - accuracy: 0.1003 - val_loss: 51.7814 - val_accuracy: 0.0979\n",
      "Epoch 2/20\n",
      "458/458 [==============================] - 30s 66ms/step - loss: 36.1470 - accuracy: 0.2697 - val_loss: 32.5720 - val_accuracy: 0.2511\n",
      "Epoch 3/20\n",
      "458/458 [==============================] - 30s 66ms/step - loss: 23.5153 - accuracy: 0.4017 - val_loss: 21.5018 - val_accuracy: 0.3835\n",
      "Epoch 4/20\n",
      "458/458 [==============================] - 30s 66ms/step - loss: 16.3966 - accuracy: 0.5025 - val_loss: 15.4748 - val_accuracy: 0.4869\n",
      "Epoch 5/20\n",
      "458/458 [==============================] - 31s 66ms/step - loss: 12.0363 - accuracy: 0.5906 - val_loss: 11.6979 - val_accuracy: 0.5722\n",
      "Epoch 6/20\n",
      "458/458 [==============================] - 30s 66ms/step - loss: 9.1691 - accuracy: 0.6546 - val_loss: 9.0900 - val_accuracy: 0.6389\n",
      "Epoch 7/20\n",
      "458/458 [==============================] - 35s 76ms/step - loss: 7.1792 - accuracy: 0.7035 - val_loss: 7.3609 - val_accuracy: 0.6904\n",
      "Epoch 8/20\n",
      "458/458 [==============================] - 35s 76ms/step - loss: 5.6794 - accuracy: 0.7505 - val_loss: 6.0575 - val_accuracy: 0.7336\n",
      "Epoch 9/20\n",
      "458/458 [==============================] - 35s 76ms/step - loss: 4.5822 - accuracy: 0.7898 - val_loss: 5.1009 - val_accuracy: 0.7626\n",
      "Epoch 10/20\n",
      "458/458 [==============================] - 30s 66ms/step - loss: 3.7382 - accuracy: 0.8188 - val_loss: 4.4177 - val_accuracy: 0.7932\n",
      "Epoch 11/20\n",
      "458/458 [==============================] - 30s 65ms/step - loss: 3.0370 - accuracy: 0.8412 - val_loss: 3.8714 - val_accuracy: 0.8129\n",
      "Epoch 12/20\n",
      "458/458 [==============================] - 35s 75ms/step - loss: 2.4254 - accuracy: 0.8648 - val_loss: 3.3831 - val_accuracy: 0.8282\n",
      "Epoch 13/20\n",
      "458/458 [==============================] - 35s 76ms/step - loss: 1.9788 - accuracy: 0.8811 - val_loss: 2.9616 - val_accuracy: 0.8479\n",
      "Epoch 14/20\n",
      "458/458 [==============================] - 30s 66ms/step - loss: 1.5861 - accuracy: 0.8998 - val_loss: 2.7780 - val_accuracy: 0.8490\n",
      "Epoch 15/20\n",
      "458/458 [==============================] - 30s 66ms/step - loss: 1.2786 - accuracy: 0.9138 - val_loss: 2.6266 - val_accuracy: 0.8534\n",
      "Epoch 16/20\n",
      "458/458 [==============================] - 30s 65ms/step - loss: 1.0217 - accuracy: 0.9245 - val_loss: 2.2974 - val_accuracy: 0.8807\n",
      "Epoch 17/20\n",
      "458/458 [==============================] - 34s 75ms/step - loss: 0.8105 - accuracy: 0.9370 - val_loss: 2.0069 - val_accuracy: 0.8911\n",
      "Epoch 18/20\n",
      "458/458 [==============================] - 30s 65ms/step - loss: 0.6342 - accuracy: 0.9453 - val_loss: 1.9913 - val_accuracy: 0.8840\n",
      "Epoch 19/20\n",
      "458/458 [==============================] - 30s 65ms/step - loss: 0.4882 - accuracy: 0.9575 - val_loss: 1.7172 - val_accuracy: 0.9054\n",
      "Epoch 20/20\n",
      "458/458 [==============================] - 30s 65ms/step - loss: 0.3802 - accuracy: 0.9616 - val_loss: 1.6879 - val_accuracy: 0.9010\n"
     ]
    }
   ],
   "source": [
    "optimizer = tf.keras.optimizers.Adam(\n",
    "    learning_rate=0.0001,\n",
    "    beta_1=0.9,\n",
    "    beta_2=0.999,\n",
    "    epsilon=1e-07\n",
    ")\n",
    "\n",
    "good_model_2_adam.compile(\n",
    "    optimizer=optimizer,\n",
    "    loss='categorical_crossentropy',\n",
    "    metrics='accuracy')\n",
    "\n",
    "h_good_2_adam = good_model_2_adam.fit(train_dataset, validation_data = val_dataset, epochs = 20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "cVam2cggmjCZ"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "cnn_transfer_learning.ipynb",
   "provenance": []
  },
  "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": "3.8.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}