pretrained_dogs_vs_cats.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: Logging before flag parsing goes to stderr.\n",
      "W0515 17:53:43.733068 140120845604608 deprecation.py:506] From /home/joel/anaconda3/envs/tensorflow-gpu-3-5-2/lib/python3.5/site-packages/tensorflow/python/ops/init_ops.py:1257: calling VarianceScaling.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Call initializer instance with the dtype argument instead of passing it to the constructor\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"vgg16\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 150, 150, 3)]     0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, 150, 150, 64)      1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, 150, 150, 64)      36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, 75, 75, 64)        0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, 75, 75, 128)       73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, 75, 75, 128)       147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, 37, 37, 128)       0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, 37, 37, 256)       295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, 37, 37, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, 37, 37, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, 18, 18, 256)       0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, 18, 18, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, 18, 18, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, 18, 18, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, 9, 9, 512)         0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, 4, 4, 512)         0         \n",
      "=================================================================\n",
      "Total params: 14,714,688\n",
      "Trainable params: 14,714,688\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "#Installing VGG16 convolutional base\n",
    "from tensorflow.keras.applications import VGG16\n",
    "\n",
    "conv_base = VGG16(weights='imagenet', include_top=False, input_shape=(150,150,3))\n",
    "\n",
    "conv_base.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2000 images belonging to 2 classes.\n",
      "Found 1000 images belonging to 2 classes.\n",
      "Found 1000 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "#Extracting features using pretrained conv base\n",
    "\n",
    "import os\n",
    "import numpy as np\n",
    "from tensorflow.keras.preprocessing.image import  ImageDataGenerator\n",
    "\n",
    "base_dir = '/home/joel/Dogs_vs_Cats_Dataset_small/'\n",
    "train_dir = os.path.join(base_dir, 'train')\n",
    "validation_dir = os.path.join(base_dir, 'validation')\n",
    "test_dir = os.path.join(base_dir, 'test')\n",
    "\n",
    "datagen = ImageDataGenerator(rescale=1./255)\n",
    "batch_size = 20\n",
    "\n",
    "def extract_features(directory, sample_count):\n",
    "    features = np.zeros(shape=(sample_count, 4, 4, 512))\n",
    "    labels = np.zeros(shape=(sample_count))\n",
    "    generator = datagen.flow_from_directory(directory, target_size=(150,150), batch_size=batch_size,\n",
    "                                           class_mode='binary')\n",
    "    i = 0\n",
    "    for inputs_batch, labels_batch in generator:\n",
    "        features_batch = conv_base.predict(inputs_batch)\n",
    "        features[i * batch_size : (i + 1) * batch_size] = features_batch #replacing the i * batch_size to (i+1) *\n",
    "                                                                         #batch_size (4,4,512)-zero arrays \n",
    "                                                                         #with the predictions\n",
    "        labels[i * batch_size : (i + 1) * batch_size] = labels_batch     #same for labels\n",
    "        i += 1\n",
    "        if i * batch_size >= sample_count:\n",
    "            break #break the indefinitely loop after every image has seen once\n",
    "    return features, labels\n",
    "\n",
    "train_features, train_labels = extract_features(train_dir, 2000)\n",
    "validation_features, validation_labels = extract_features(validation_dir, 1000)\n",
    "test_features, test_labels = extract_features(test_dir,1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_features = np.reshape(train_features, (2000, 4*4*512))\n",
    "validation_features = np.reshape(validation_features, (1000, 4*4*512))\n",
    "test_features = np.reshape(test_features, (1000, 4*4*512))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 2000 samples, validate on 1000 samples\n",
      "Epoch 1/30\n",
      "2000/2000 [==============================] - 1s 371us/sample - loss: 0.6272 - acc: 0.6500 - val_loss: 0.4646 - val_acc: 0.8220\n",
      "Epoch 2/30\n",
      "2000/2000 [==============================] - 0s 215us/sample - loss: 0.4379 - acc: 0.8075 - val_loss: 0.3740 - val_acc: 0.8600\n",
      "Epoch 3/30\n",
      "2000/2000 [==============================] - 0s 212us/sample - loss: 0.3780 - acc: 0.8405 - val_loss: 0.3313 - val_acc: 0.8720\n",
      "Epoch 4/30\n",
      "2000/2000 [==============================] - 0s 223us/sample - loss: 0.3202 - acc: 0.8600 - val_loss: 0.3128 - val_acc: 0.8810\n",
      "Epoch 5/30\n",
      "2000/2000 [==============================] - 0s 216us/sample - loss: 0.2899 - acc: 0.8830 - val_loss: 0.2880 - val_acc: 0.8860\n",
      "Epoch 6/30\n",
      "2000/2000 [==============================] - 0s 218us/sample - loss: 0.2725 - acc: 0.8975 - val_loss: 0.2793 - val_acc: 0.8850\n",
      "Epoch 7/30\n",
      "2000/2000 [==============================] - 0s 212us/sample - loss: 0.2525 - acc: 0.8980 - val_loss: 0.2716 - val_acc: 0.8890\n",
      "Epoch 8/30\n",
      "2000/2000 [==============================] - 0s 220us/sample - loss: 0.2362 - acc: 0.9060 - val_loss: 0.2664 - val_acc: 0.8880\n",
      "Epoch 9/30\n",
      "2000/2000 [==============================] - 0s 214us/sample - loss: 0.2204 - acc: 0.9105 - val_loss: 0.2692 - val_acc: 0.8800\n",
      "Epoch 10/30\n",
      "2000/2000 [==============================] - 0s 211us/sample - loss: 0.2110 - acc: 0.9260 - val_loss: 0.2494 - val_acc: 0.8960\n",
      "Epoch 11/30\n",
      "2000/2000 [==============================] - 0s 212us/sample - loss: 0.2010 - acc: 0.9260 - val_loss: 0.2489 - val_acc: 0.8910\n",
      "Epoch 12/30\n",
      "2000/2000 [==============================] - 0s 214us/sample - loss: 0.1940 - acc: 0.9295 - val_loss: 0.2440 - val_acc: 0.9010\n",
      "Epoch 13/30\n",
      "2000/2000 [==============================] - 0s 213us/sample - loss: 0.1825 - acc: 0.9385 - val_loss: 0.2545 - val_acc: 0.8880\n",
      "Epoch 14/30\n",
      "2000/2000 [==============================] - 0s 211us/sample - loss: 0.1737 - acc: 0.9355 - val_loss: 0.2556 - val_acc: 0.8870\n",
      "Epoch 15/30\n",
      "2000/2000 [==============================] - 0s 216us/sample - loss: 0.1699 - acc: 0.9375 - val_loss: 0.2395 - val_acc: 0.9030\n",
      "Epoch 16/30\n",
      "2000/2000 [==============================] - 0s 212us/sample - loss: 0.1575 - acc: 0.9390 - val_loss: 0.2408 - val_acc: 0.8980\n",
      "Epoch 17/30\n",
      "2000/2000 [==============================] - 0s 212us/sample - loss: 0.1528 - acc: 0.9475 - val_loss: 0.2403 - val_acc: 0.8950\n",
      "Epoch 18/30\n",
      "2000/2000 [==============================] - 0s 211us/sample - loss: 0.1525 - acc: 0.9495 - val_loss: 0.2353 - val_acc: 0.9050\n",
      "Epoch 19/30\n",
      "2000/2000 [==============================] - 0s 214us/sample - loss: 0.1419 - acc: 0.9535 - val_loss: 0.2445 - val_acc: 0.8990\n",
      "Epoch 20/30\n",
      "2000/2000 [==============================] - 0s 215us/sample - loss: 0.1406 - acc: 0.9510 - val_loss: 0.2375 - val_acc: 0.9000\n",
      "Epoch 21/30\n",
      "2000/2000 [==============================] - 0s 213us/sample - loss: 0.1309 - acc: 0.9535 - val_loss: 0.2332 - val_acc: 0.9060\n",
      "Epoch 22/30\n",
      "2000/2000 [==============================] - 0s 215us/sample - loss: 0.1216 - acc: 0.9620 - val_loss: 0.2361 - val_acc: 0.8990\n",
      "Epoch 23/30\n",
      "2000/2000 [==============================] - 0s 217us/sample - loss: 0.1185 - acc: 0.9615 - val_loss: 0.2379 - val_acc: 0.9000\n",
      "Epoch 24/30\n",
      "2000/2000 [==============================] - 0s 215us/sample - loss: 0.1163 - acc: 0.9645 - val_loss: 0.2358 - val_acc: 0.9010\n",
      "Epoch 25/30\n",
      "2000/2000 [==============================] - 0s 215us/sample - loss: 0.1116 - acc: 0.9640 - val_loss: 0.2408 - val_acc: 0.8970\n",
      "Epoch 26/30\n",
      "2000/2000 [==============================] - 0s 216us/sample - loss: 0.1099 - acc: 0.9630 - val_loss: 0.2357 - val_acc: 0.9010\n",
      "Epoch 27/30\n",
      "2000/2000 [==============================] - 0s 211us/sample - loss: 0.1016 - acc: 0.9670 - val_loss: 0.2370 - val_acc: 0.9000\n",
      "Epoch 28/30\n",
      "2000/2000 [==============================] - 0s 213us/sample - loss: 0.1013 - acc: 0.9670 - val_loss: 0.2427 - val_acc: 0.9010\n",
      "Epoch 29/30\n",
      "2000/2000 [==============================] - 0s 218us/sample - loss: 0.0956 - acc: 0.9715 - val_loss: 0.2399 - val_acc: 0.9040\n",
      "Epoch 30/30\n",
      "2000/2000 [==============================] - 0s 211us/sample - loss: 0.0858 - acc: 0.9775 - val_loss: 0.2384 - val_acc: 0.9000\n"
     ]
    }
   ],
   "source": [
    "#Define densely connected classifier with dropout and train it\n",
    "\n",
    "from tensorflow.keras import models\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras import optimizers\n",
    "\n",
    "model = models.Sequential()\n",
    "model.add(layers.Dense(256, activation='relu', input_dim=4*4*512))\n",
    "model.add(layers.Dropout(0.5))\n",
    "model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "model.compile(optimizer=optimizers.RMSprop(lr=2e-5), loss='binary_crossentropy', metrics=['acc'])\n",
    "\n",
    "history = model.fit(train_features, train_labels, epochs=30,batch_size=20,\n",
    "                    validation_data=(validation_features,validation_labels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Plotting test and validation accuracy and loss\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(1, len(acc)+1)\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Adding the conv_base to a Sequential model for running the inputs\n",
    "#end-to-end\n",
    "\n",
    "from tensorflow.keras import models\n",
    "from tensorflow.keras import layers\n",
    "\n",
    "model = models.Sequential()\n",
    "model.add(conv_base)\n",
    "model.add(layers.Flatten())\n",
    "model.add(layers.Dense(256, activation='relu'))\n",
    "model.add(layers.Dense(1, activation='sigmoid'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "vgg16 (Model)                (None, 4, 4, 512)         14714688  \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 8192)              0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 256)               2097408   \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 1)                 257       \n",
      "=================================================================\n",
      "Total params: 16,812,353\n",
      "Trainable params: 16,812,353\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "This is the number of trainable weights before freezing the conv base: 30\n",
      "This is the number of trainable weights after freezing the conv base: 4\n"
     ]
    }
   ],
   "source": [
    "#before compiling and trianing the model, we have to freeze\n",
    "#the conv base (since we don't want to modify the previous learned\n",
    "#representation during training)\n",
    "\n",
    "print('This is the number of trainable weights '\n",
    "      'before freezing the conv base:', len(model.trainable_weights))\n",
    "\n",
    "conv_base.trainable=False\n",
    "\n",
    "print('This is the number of trainable weights '\n",
    "      'after freezing the conv base:', len(model.trainable_weights))\n",
    "\n",
    "#The trainable weights are indeed the Weight matrices and the bias vector"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2000 images belonging to 2 classes.\n",
      "Found 1000 images belonging to 2 classes.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "100/100 - 16s - loss: 0.5874 - acc: 0.6980 - val_loss: 0.4329 - val_acc: 0.8410\n",
      "Epoch 2/30\n",
      "100/100 - 15s - loss: 0.4797 - acc: 0.7930 - val_loss: 0.3640 - val_acc: 0.8690\n",
      "Epoch 3/30\n",
      "100/100 - 15s - loss: 0.4269 - acc: 0.8190 - val_loss: 0.3169 - val_acc: 0.8860\n",
      "Epoch 4/30\n",
      "100/100 - 15s - loss: 0.4084 - acc: 0.8175 - val_loss: 0.2971 - val_acc: 0.8810\n",
      "Epoch 5/30\n",
      "100/100 - 15s - loss: 0.3830 - acc: 0.8330 - val_loss: 0.2832 - val_acc: 0.8860\n",
      "Epoch 6/30\n",
      "100/100 - 15s - loss: 0.3729 - acc: 0.8360 - val_loss: 0.2732 - val_acc: 0.8900\n",
      "Epoch 7/30\n",
      "100/100 - 15s - loss: 0.3666 - acc: 0.8370 - val_loss: 0.2670 - val_acc: 0.8930\n",
      "Epoch 8/30\n",
      "100/100 - 15s - loss: 0.3516 - acc: 0.8390 - val_loss: 0.2659 - val_acc: 0.8980\n",
      "Epoch 9/30\n",
      "100/100 - 15s - loss: 0.3488 - acc: 0.8445 - val_loss: 0.2563 - val_acc: 0.9030\n",
      "Epoch 10/30\n",
      "100/100 - 15s - loss: 0.3422 - acc: 0.8495 - val_loss: 0.2520 - val_acc: 0.8980\n",
      "Epoch 11/30\n",
      "100/100 - 15s - loss: 0.3372 - acc: 0.8500 - val_loss: 0.2505 - val_acc: 0.9010\n",
      "Epoch 12/30\n",
      "100/100 - 15s - loss: 0.3132 - acc: 0.8650 - val_loss: 0.2476 - val_acc: 0.9030\n",
      "Epoch 13/30\n",
      "100/100 - 15s - loss: 0.3236 - acc: 0.8660 - val_loss: 0.2421 - val_acc: 0.9090\n",
      "Epoch 14/30\n",
      "100/100 - 15s - loss: 0.3096 - acc: 0.8690 - val_loss: 0.2413 - val_acc: 0.9090\n",
      "Epoch 15/30\n",
      "100/100 - 15s - loss: 0.3217 - acc: 0.8510 - val_loss: 0.2594 - val_acc: 0.8890\n",
      "Epoch 16/30\n",
      "100/100 - 15s - loss: 0.3139 - acc: 0.8645 - val_loss: 0.2385 - val_acc: 0.9120\n",
      "Epoch 17/30\n",
      "100/100 - 15s - loss: 0.3029 - acc: 0.8715 - val_loss: 0.2451 - val_acc: 0.9020\n",
      "Epoch 18/30\n",
      "100/100 - 15s - loss: 0.3045 - acc: 0.8665 - val_loss: 0.2396 - val_acc: 0.8990\n",
      "Epoch 19/30\n",
      "100/100 - 15s - loss: 0.2992 - acc: 0.8645 - val_loss: 0.2357 - val_acc: 0.9080\n",
      "Epoch 20/30\n",
      "100/100 - 15s - loss: 0.2948 - acc: 0.8745 - val_loss: 0.2368 - val_acc: 0.9050\n",
      "Epoch 21/30\n",
      "100/100 - 15s - loss: 0.2853 - acc: 0.8790 - val_loss: 0.2401 - val_acc: 0.9040\n",
      "Epoch 22/30\n",
      "100/100 - 15s - loss: 0.2984 - acc: 0.8730 - val_loss: 0.2348 - val_acc: 0.9080\n",
      "Epoch 23/30\n",
      "100/100 - 15s - loss: 0.2921 - acc: 0.8720 - val_loss: 0.2407 - val_acc: 0.9070\n",
      "Epoch 24/30\n",
      "100/100 - 15s - loss: 0.2939 - acc: 0.8750 - val_loss: 0.2482 - val_acc: 0.8940\n",
      "Epoch 25/30\n",
      "100/100 - 15s - loss: 0.3030 - acc: 0.8665 - val_loss: 0.2483 - val_acc: 0.8960\n",
      "Epoch 26/30\n",
      "100/100 - 15s - loss: 0.2797 - acc: 0.8800 - val_loss: 0.2400 - val_acc: 0.9060\n",
      "Epoch 27/30\n",
      "100/100 - 15s - loss: 0.2848 - acc: 0.8790 - val_loss: 0.2438 - val_acc: 0.8990\n",
      "Epoch 28/30\n",
      "100/100 - 15s - loss: 0.2871 - acc: 0.8780 - val_loss: 0.2339 - val_acc: 0.9060\n",
      "Epoch 29/30\n",
      "100/100 - 15s - loss: 0.2788 - acc: 0.8790 - val_loss: 0.2347 - val_acc: 0.9060\n",
      "Epoch 30/30\n",
      "100/100 - 15s - loss: 0.2860 - acc: 0.8710 - val_loss: 0.2330 - val_acc: 0.9110\n"
     ]
    }
   ],
   "source": [
    "#training the model end-to-end with frozen conv-base\n",
    "\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "train_datagen = ImageDataGenerator(\n",
    "      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",
    "      fill_mode='nearest')\n",
    "\n",
    "# Note that the validation data should not be augmented!\n",
    "test_datagen = ImageDataGenerator(rescale=1./255)\n",
    "\n",
    "train_generator = train_datagen.flow_from_directory(\n",
    "        train_dir,\n",
    "        target_size=(150, 150),\n",
    "        batch_size=20,\n",
    "        class_mode='binary')\n",
    "\n",
    "validation_generator = test_datagen.flow_from_directory(\n",
    "        validation_dir,\n",
    "        target_size=(150, 150),\n",
    "        batch_size=20,\n",
    "        class_mode='binary')\n",
    "\n",
    "model.compile(loss='binary_crossentropy',\n",
    "              optimizer=optimizers.RMSprop(lr=2e-5),\n",
    "              metrics=['acc'])\n",
    "\n",
    "history = model.fit_generator(\n",
    "      train_generator,\n",
    "      steps_per_epoch=100,\n",
    "      epochs=30,\n",
    "      validation_data=validation_generator,\n",
    "      validation_steps=50,\n",
    "      verbose=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Plotting the accuracy and loss for the model with frozen conv base and\n",
    "#data augmentation\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(1, len(acc)+1)\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"vgg16\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 150, 150, 3)]     0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, 150, 150, 64)      1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, 150, 150, 64)      36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, 75, 75, 64)        0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, 75, 75, 128)       73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, 75, 75, 128)       147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, 37, 37, 128)       0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, 37, 37, 256)       295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, 37, 37, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, 37, 37, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, 18, 18, 256)       0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, 18, 18, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, 18, 18, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, 18, 18, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, 9, 9, 512)         0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, 4, 4, 512)         0         \n",
      "=================================================================\n",
      "Total params: 14,714,688\n",
      "Trainable params: 0\n",
      "Non-trainable params: 14,714,688\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "conv_base.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "#In order to fine-tune the model we unfreeze the last three convolutional layers\n",
    "\n",
    "conv_base.trainable = True\n",
    "\n",
    "set_trainable = False\n",
    "for layer in conv_base.layers:\n",
    "    if layer.name == 'block5_conv1':\n",
    "        set_trainable = True\n",
    "    if set_trainable:\n",
    "        layer.trainable = True\n",
    "    else:\n",
    "        layer.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "100/100 [==============================] - 16s 163ms/step - loss: 0.2808 - acc: 0.8740 - val_loss: 0.2208 - val_acc: 0.9140\n",
      "Epoch 2/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.2418 - acc: 0.8905 - val_loss: 0.2090 - val_acc: 0.9210\n",
      "Epoch 3/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.2485 - acc: 0.8885 - val_loss: 0.1950 - val_acc: 0.9230\n",
      "Epoch 4/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.2329 - acc: 0.9045 - val_loss: 0.2071 - val_acc: 0.9240\n",
      "Epoch 5/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.2147 - acc: 0.9145 - val_loss: 0.1891 - val_acc: 0.9300\n",
      "Epoch 6/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.1892 - acc: 0.9195 - val_loss: 0.1852 - val_acc: 0.9280\n",
      "Epoch 7/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.1727 - acc: 0.9255 - val_loss: 0.1944 - val_acc: 0.9320\n",
      "Epoch 8/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.1857 - acc: 0.9220 - val_loss: 0.2156 - val_acc: 0.9200\n",
      "Epoch 9/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.1607 - acc: 0.9340 - val_loss: 0.1722 - val_acc: 0.9360\n",
      "Epoch 10/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.1548 - acc: 0.9385 - val_loss: 0.1775 - val_acc: 0.9290\n",
      "Epoch 11/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.1380 - acc: 0.9440 - val_loss: 0.1828 - val_acc: 0.9300\n",
      "Epoch 12/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.1442 - acc: 0.9435 - val_loss: 0.1905 - val_acc: 0.9280\n",
      "Epoch 13/100\n",
      "100/100 [==============================] - 15s 154ms/step - loss: 0.1394 - acc: 0.9440 - val_loss: 0.2038 - val_acc: 0.9220\n",
      "Epoch 14/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.1194 - acc: 0.9520 - val_loss: 0.1878 - val_acc: 0.9340\n",
      "Epoch 15/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.1270 - acc: 0.9475 - val_loss: 0.1781 - val_acc: 0.9330\n",
      "Epoch 16/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.1329 - acc: 0.9440 - val_loss: 0.1787 - val_acc: 0.9330\n",
      "Epoch 17/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.1069 - acc: 0.9585 - val_loss: 0.2218 - val_acc: 0.9170\n",
      "Epoch 18/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.1110 - acc: 0.9560 - val_loss: 0.1792 - val_acc: 0.9370\n",
      "Epoch 19/100\n",
      "100/100 [==============================] - 15s 154ms/step - loss: 0.1105 - acc: 0.9600 - val_loss: 0.1771 - val_acc: 0.9350\n",
      "Epoch 20/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0957 - acc: 0.9620 - val_loss: 0.2148 - val_acc: 0.9260\n",
      "Epoch 21/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.1001 - acc: 0.9620 - val_loss: 0.2276 - val_acc: 0.9240\n",
      "Epoch 22/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0901 - acc: 0.9605 - val_loss: 0.1886 - val_acc: 0.9330\n",
      "Epoch 23/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0993 - acc: 0.9585 - val_loss: 0.1911 - val_acc: 0.9290\n",
      "Epoch 24/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0802 - acc: 0.9675 - val_loss: 0.1948 - val_acc: 0.9320\n",
      "Epoch 25/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0724 - acc: 0.9755 - val_loss: 0.2623 - val_acc: 0.9210\n",
      "Epoch 26/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0762 - acc: 0.9715 - val_loss: 0.1965 - val_acc: 0.9320\n",
      "Epoch 27/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0699 - acc: 0.9740 - val_loss: 0.3533 - val_acc: 0.8990\n",
      "Epoch 28/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0742 - acc: 0.9720 - val_loss: 0.1774 - val_acc: 0.9310\n",
      "Epoch 29/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0756 - acc: 0.9725 - val_loss: 0.2181 - val_acc: 0.9260\n",
      "Epoch 30/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0565 - acc: 0.9775 - val_loss: 0.2011 - val_acc: 0.9330\n",
      "Epoch 31/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0714 - acc: 0.9755 - val_loss: 0.1877 - val_acc: 0.9390\n",
      "Epoch 32/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0513 - acc: 0.9810 - val_loss: 0.2004 - val_acc: 0.9370\n",
      "Epoch 33/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0670 - acc: 0.9725 - val_loss: 0.2553 - val_acc: 0.9270\n",
      "Epoch 34/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0592 - acc: 0.9785 - val_loss: 0.1783 - val_acc: 0.9420\n",
      "Epoch 35/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0603 - acc: 0.9740 - val_loss: 0.2188 - val_acc: 0.9300\n",
      "Epoch 36/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0637 - acc: 0.9770 - val_loss: 0.2589 - val_acc: 0.9330\n",
      "Epoch 37/100\n",
      "100/100 [==============================] - 15s 154ms/step - loss: 0.0579 - acc: 0.9765 - val_loss: 0.2032 - val_acc: 0.9460\n",
      "Epoch 38/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0612 - acc: 0.9755 - val_loss: 0.2161 - val_acc: 0.9310\n",
      "Epoch 39/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0555 - acc: 0.9775 - val_loss: 0.2213 - val_acc: 0.9330\n",
      "Epoch 40/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0483 - acc: 0.9825 - val_loss: 0.2458 - val_acc: 0.9410\n",
      "Epoch 41/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0536 - acc: 0.9825 - val_loss: 0.2857 - val_acc: 0.9200\n",
      "Epoch 42/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0410 - acc: 0.9820 - val_loss: 0.3835 - val_acc: 0.9080\n",
      "Epoch 43/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0496 - acc: 0.9825 - val_loss: 0.2156 - val_acc: 0.9330\n",
      "Epoch 44/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0374 - acc: 0.9880 - val_loss: 0.3323 - val_acc: 0.9150\n",
      "Epoch 45/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0488 - acc: 0.9825 - val_loss: 0.2124 - val_acc: 0.9370\n",
      "Epoch 46/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0514 - acc: 0.9800 - val_loss: 0.2191 - val_acc: 0.9340\n",
      "Epoch 47/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0503 - acc: 0.9815 - val_loss: 0.3016 - val_acc: 0.9250\n",
      "Epoch 48/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0408 - acc: 0.9845 - val_loss: 0.2221 - val_acc: 0.9400\n",
      "Epoch 49/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0431 - acc: 0.9865 - val_loss: 0.2232 - val_acc: 0.9380\n",
      "Epoch 50/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0399 - acc: 0.9895 - val_loss: 0.2072 - val_acc: 0.9410\n",
      "Epoch 51/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0433 - acc: 0.9860 - val_loss: 0.2149 - val_acc: 0.9400\n",
      "Epoch 52/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0393 - acc: 0.9870 - val_loss: 0.1995 - val_acc: 0.9360\n",
      "Epoch 53/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0353 - acc: 0.9890 - val_loss: 0.2275 - val_acc: 0.9420\n",
      "Epoch 54/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0386 - acc: 0.9870 - val_loss: 0.1854 - val_acc: 0.9360\n",
      "Epoch 55/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0332 - acc: 0.9865 - val_loss: 0.3522 - val_acc: 0.9230\n",
      "Epoch 56/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0366 - acc: 0.9860 - val_loss: 0.2294 - val_acc: 0.9320\n",
      "Epoch 57/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0312 - acc: 0.9875 - val_loss: 0.2190 - val_acc: 0.9390\n",
      "Epoch 58/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0272 - acc: 0.9900 - val_loss: 0.2158 - val_acc: 0.9380\n",
      "Epoch 59/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0336 - acc: 0.9900 - val_loss: 0.2498 - val_acc: 0.9380\n",
      "Epoch 60/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0401 - acc: 0.9875 - val_loss: 0.2505 - val_acc: 0.9390\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 61/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0257 - acc: 0.9900 - val_loss: 0.2641 - val_acc: 0.9370\n",
      "Epoch 62/100\n",
      "100/100 [==============================] - 15s 154ms/step - loss: 0.0339 - acc: 0.9855 - val_loss: 0.2501 - val_acc: 0.9410\n",
      "Epoch 63/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0342 - acc: 0.9860 - val_loss: 0.2183 - val_acc: 0.9430\n",
      "Epoch 64/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0275 - acc: 0.9905 - val_loss: 0.2335 - val_acc: 0.9330\n",
      "Epoch 65/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0371 - acc: 0.9870 - val_loss: 0.2311 - val_acc: 0.9390\n",
      "Epoch 66/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0223 - acc: 0.9915 - val_loss: 0.2516 - val_acc: 0.9350\n",
      "Epoch 67/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0400 - acc: 0.9880 - val_loss: 0.2212 - val_acc: 0.9430\n",
      "Epoch 68/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0244 - acc: 0.9910 - val_loss: 0.2306 - val_acc: 0.9410\n",
      "Epoch 69/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0267 - acc: 0.9915 - val_loss: 0.2627 - val_acc: 0.9400\n",
      "Epoch 70/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0249 - acc: 0.9900 - val_loss: 0.3106 - val_acc: 0.9280\n",
      "Epoch 71/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0327 - acc: 0.9895 - val_loss: 0.2214 - val_acc: 0.9400\n",
      "Epoch 72/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0234 - acc: 0.9935 - val_loss: 0.2104 - val_acc: 0.9400\n",
      "Epoch 73/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0339 - acc: 0.9880 - val_loss: 0.2341 - val_acc: 0.9430\n",
      "Epoch 74/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0232 - acc: 0.9915 - val_loss: 0.2585 - val_acc: 0.9390\n",
      "Epoch 75/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0268 - acc: 0.9910 - val_loss: 0.2463 - val_acc: 0.9340\n",
      "Epoch 76/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0277 - acc: 0.9905 - val_loss: 0.2406 - val_acc: 0.9350\n",
      "Epoch 77/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0171 - acc: 0.9940 - val_loss: 0.2384 - val_acc: 0.9430\n",
      "Epoch 78/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0244 - acc: 0.9920 - val_loss: 0.3449 - val_acc: 0.9350\n",
      "Epoch 79/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0243 - acc: 0.9925 - val_loss: 0.5464 - val_acc: 0.9060\n",
      "Epoch 80/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0187 - acc: 0.9935 - val_loss: 0.2558 - val_acc: 0.9350\n",
      "Epoch 81/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0203 - acc: 0.9930 - val_loss: 0.3696 - val_acc: 0.9270\n",
      "Epoch 82/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0138 - acc: 0.9950 - val_loss: 0.2878 - val_acc: 0.9370\n",
      "Epoch 83/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0248 - acc: 0.9930 - val_loss: 0.2635 - val_acc: 0.9400\n",
      "Epoch 84/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0295 - acc: 0.9910 - val_loss: 0.2686 - val_acc: 0.9340\n",
      "Epoch 85/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0188 - acc: 0.9950 - val_loss: 0.4070 - val_acc: 0.9240\n",
      "Epoch 86/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0139 - acc: 0.9955 - val_loss: 0.2969 - val_acc: 0.9340\n",
      "Epoch 87/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0229 - acc: 0.9930 - val_loss: 0.2914 - val_acc: 0.9320\n",
      "Epoch 88/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0178 - acc: 0.9925 - val_loss: 0.2276 - val_acc: 0.9420\n",
      "Epoch 89/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0169 - acc: 0.9945 - val_loss: 0.3010 - val_acc: 0.9350\n",
      "Epoch 90/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0203 - acc: 0.9925 - val_loss: 0.2959 - val_acc: 0.9360\n",
      "Epoch 91/100\n",
      "100/100 [==============================] - 15s 150ms/step - loss: 0.0213 - acc: 0.9920 - val_loss: 0.3274 - val_acc: 0.9360\n",
      "Epoch 92/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0161 - acc: 0.9930 - val_loss: 0.2776 - val_acc: 0.9340\n",
      "Epoch 93/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0252 - acc: 0.9900 - val_loss: 0.2694 - val_acc: 0.9420\n",
      "Epoch 94/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0176 - acc: 0.9955 - val_loss: 0.2252 - val_acc: 0.9390\n",
      "Epoch 95/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0253 - acc: 0.9915 - val_loss: 0.2546 - val_acc: 0.9390\n",
      "Epoch 96/100\n",
      "100/100 [==============================] - 15s 152ms/step - loss: 0.0188 - acc: 0.9925 - val_loss: 0.2704 - val_acc: 0.9360\n",
      "Epoch 97/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0183 - acc: 0.9930 - val_loss: 0.4467 - val_acc: 0.9240\n",
      "Epoch 98/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0191 - acc: 0.9930 - val_loss: 0.2680 - val_acc: 0.9390\n",
      "Epoch 99/100\n",
      "100/100 [==============================] - 15s 153ms/step - loss: 0.0122 - acc: 0.9960 - val_loss: 0.3749 - val_acc: 0.9310\n",
      "Epoch 100/100\n",
      "100/100 [==============================] - 15s 151ms/step - loss: 0.0156 - acc: 0.9950 - val_loss: 0.2933 - val_acc: 0.9420\n"
     ]
    }
   ],
   "source": [
    "#fine-tuning the model with a very low learning rate in order to keep the \n",
    "#limit the magnitude of the modifications to the representations\n",
    "\n",
    "model.compile(loss='binary_crossentropy', optimizer = optimizers.RMSprop(lr=1e-5),metrics=['acc'])\n",
    "\n",
    "history = model.fit_generator(train_generator, steps_per_epoch=100,epochs=100,validation_data=validation_generator,validation_steps=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Plotting the accuracy and loss for the model with unfrozen last three conv layers\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(1, len(acc)+1)\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#In order to make the plot more readable, we smooth the points by their exp moving averages\n",
    "\n",
    "def smooth_curve(points, factor = 0.8):\n",
    "    smoothed_points=[]\n",
    "    for point in points:\n",
    "        if smoothed_points:\n",
    "            previous = smoothed_points[-1]\n",
    "            smoothed_points.append(previous * factor + point * (1-factor))\n",
    "        else:\n",
    "            smoothed_points.append(point)\n",
    "    return smoothed_points\n",
    "\n",
    "plt.plot(epochs, smooth_curve(acc), 'bo', label='Smoothed training acc')\n",
    "plt.plot(epochs, smooth_curve(val_acc), 'b', label='Smoothed validation acc')\n",
    "plt.title('Smoothed training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, smooth_curve(loss), 'bo', label='Smoothed training loss')\n",
    "plt.plot(epochs, smooth_curve(val_loss), 'b', label='Smoothed validation loss')\n",
    "plt.title('Smoothed training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 1000 images belonging to 2 classes.\n",
      "test acc: 0.942\n"
     ]
    }
   ],
   "source": [
    "#Evaluate the model on the test set\n",
    "\n",
    "test_generator = test_datagen.flow_from_directory(test_dir, target_size=(150,150), batch_size=20, class_mode='binary')\n",
    "\n",
    "test_loss, test_acc = model.evaluate_generator(test_generator, steps=50)\n",
    "print('test acc:', test_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}