tl_vgg16_cnn.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_10\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_10 (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",
      "flatten_10 (Flatten)         (None, 8192)              0         \n",
      "=================================================================\n",
      "Total params: 14,714,688\n",
      "Trainable params: 0\n",
      "Non-trainable params: 14,714,688\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/ravi/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:11: UserWarning: Update your `Model` call to the Keras 2 API: `Model(Tensor(\"in..., outputs=Tensor(\"fl...)`\n",
      "  # This is added back by InteractiveShellApp.init_path()\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "np.random.seed(0)\n",
    "IMG_WIDTH=150\n",
    "IMG_HEIGHT=150\n",
    "from keras.applications.vgg16 import VGG16\n",
    "from keras.models import Model\n",
    "import keras\n",
    "restnet = VGG16(include_top=False, weights='imagenet', input_shape=(IMG_HEIGHT,IMG_WIDTH,3))\n",
    "output = restnet.layers[-1].output\n",
    "output = keras.layers.Flatten()(output)\n",
    "restnet = Model(restnet.input, output=output)\n",
    "for layer in restnet.layers:\n",
    "    layer.trainable = False\n",
    "restnet.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_3\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "model_10 (Model)             (None, 8192)              14714688  \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 256)               2097408   \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 1)                 257       \n",
      "=================================================================\n",
      "Total params: 16,812,353\n",
      "Trainable params: 2,097,665\n",
      "Non-trainable params: 14,714,688\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout, InputLayer\n",
    "from keras.models import Sequential\n",
    "from keras import optimizers\n",
    "input_shape=(IMG_HEIGHT,IMG_WIDTH,3)\n",
    "model = Sequential()\n",
    "model.add(restnet)\n",
    "model.add(Dense(256, activation='relu', input_dim=input_shape))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(1, activation='sigmoid'))\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 3998 images belonging to 2 classes.\n",
      "Found 800 images belonging to 2 classes.\n",
      "Epoch 1/100\n",
      "49/49 [==============================] - 511s 10s/step - loss: 0.4206 - accuracy: 0.8299 - val_loss: 0.3432 - val_accuracy: 0.8780\n",
      "Epoch 2/100\n",
      "49/49 [==============================] - 530s 11s/step - loss: 0.3560 - accuracy: 0.8657 - val_loss: 0.3062 - val_accuracy: 0.9030\n",
      "Epoch 3/100\n",
      "49/49 [==============================] - 530s 11s/step - loss: 0.3499 - accuracy: 0.8601 - val_loss: 0.4048 - val_accuracy: 0.8820\n",
      "Epoch 4/100\n",
      "49/49 [==============================] - 536s 11s/step - loss: 0.3392 - accuracy: 0.8666 - val_loss: 0.4095 - val_accuracy: 0.8780\n",
      "Epoch 5/100\n",
      "49/49 [==============================] - 533s 11s/step - loss: 0.3247 - accuracy: 0.8744 - val_loss: 0.3029 - val_accuracy: 0.8880\n",
      "Epoch 6/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.3173 - accuracy: 0.8770 - val_loss: 0.4023 - val_accuracy: 0.9040\n",
      "Epoch 7/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.3066 - accuracy: 0.8871 - val_loss: 0.3662 - val_accuracy: 0.8840\n",
      "Epoch 8/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.3066 - accuracy: 0.8828 - val_loss: 0.3147 - val_accuracy: 0.9080\n",
      "Epoch 9/100\n",
      "49/49 [==============================] - 532s 11s/step - loss: 0.2980 - accuracy: 0.8893 - val_loss: 0.2500 - val_accuracy: 0.9070\n",
      "Epoch 10/100\n",
      "49/49 [==============================] - 534s 11s/step - loss: 0.2928 - accuracy: 0.8855 - val_loss: 0.2321 - val_accuracy: 0.9060\n",
      "Epoch 11/100\n",
      "49/49 [==============================] - 534s 11s/step - loss: 0.2968 - accuracy: 0.8873 - val_loss: 0.2765 - val_accuracy: 0.9130\n",
      "Epoch 12/100\n",
      "49/49 [==============================] - 532s 11s/step - loss: 0.2901 - accuracy: 0.8906 - val_loss: 0.3494 - val_accuracy: 0.9100\n",
      "Epoch 13/100\n",
      "49/49 [==============================] - 529s 11s/step - loss: 0.2974 - accuracy: 0.8871 - val_loss: 0.2324 - val_accuracy: 0.9100\n",
      "Epoch 14/100\n",
      "49/49 [==============================] - 533s 11s/step - loss: 0.2809 - accuracy: 0.8926 - val_loss: 0.2611 - val_accuracy: 0.9160\n",
      "Epoch 15/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2777 - accuracy: 0.8969 - val_loss: 0.2501 - val_accuracy: 0.9170\n",
      "Epoch 16/100\n",
      "49/49 [==============================] - 532s 11s/step - loss: 0.2796 - accuracy: 0.8947 - val_loss: 0.3681 - val_accuracy: 0.9210\n",
      "Epoch 17/100\n",
      "49/49 [==============================] - 537s 11s/step - loss: 0.2711 - accuracy: 0.8924 - val_loss: 0.2493 - val_accuracy: 0.9200\n",
      "Epoch 18/100\n",
      "49/49 [==============================] - 586s 12s/step - loss: 0.2775 - accuracy: 0.8944 - val_loss: 0.2360 - val_accuracy: 0.9170\n",
      "Epoch 19/100\n",
      "49/49 [==============================] - 532s 11s/step - loss: 0.2745 - accuracy: 0.8932 - val_loss: 0.2413 - val_accuracy: 0.9190\n",
      "Epoch 20/100\n",
      "49/49 [==============================] - 532s 11s/step - loss: 0.2777 - accuracy: 0.8953 - val_loss: 0.1884 - val_accuracy: 0.9100\n",
      "Epoch 21/100\n",
      "49/49 [==============================] - 530s 11s/step - loss: 0.2672 - accuracy: 0.9000 - val_loss: 0.2292 - val_accuracy: 0.9210\n",
      "Epoch 22/100\n",
      "49/49 [==============================] - 528s 11s/step - loss: 0.2647 - accuracy: 0.8987 - val_loss: 0.2266 - val_accuracy: 0.9160\n",
      "Epoch 23/100\n",
      "49/49 [==============================] - 533s 11s/step - loss: 0.2760 - accuracy: 0.8934 - val_loss: 0.1545 - val_accuracy: 0.9380\n",
      "Epoch 24/100\n",
      "49/49 [==============================] - 539s 11s/step - loss: 0.2651 - accuracy: 0.8982 - val_loss: 0.2094 - val_accuracy: 0.9060\n",
      "Epoch 25/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2609 - accuracy: 0.9022 - val_loss: 0.3429 - val_accuracy: 0.9180\n",
      "Epoch 26/100\n",
      "49/49 [==============================] - 529s 11s/step - loss: 0.2672 - accuracy: 0.8928 - val_loss: 0.2025 - val_accuracy: 0.9310\n",
      "Epoch 27/100\n",
      "49/49 [==============================] - 532s 11s/step - loss: 0.2609 - accuracy: 0.9007 - val_loss: 0.2292 - val_accuracy: 0.9000\n",
      "Epoch 28/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2580 - accuracy: 0.9016 - val_loss: 0.2455 - val_accuracy: 0.9060\n",
      "Epoch 29/100\n",
      "49/49 [==============================] - 530s 11s/step - loss: 0.2591 - accuracy: 0.9018 - val_loss: 0.3345 - val_accuracy: 0.9190\n",
      "Epoch 30/100\n",
      "49/49 [==============================] - 532s 11s/step - loss: 0.2641 - accuracy: 0.8993 - val_loss: 0.2176 - val_accuracy: 0.9200\n",
      "Epoch 31/100\n",
      "49/49 [==============================] - 533s 11s/step - loss: 0.2540 - accuracy: 0.9024 - val_loss: 0.3152 - val_accuracy: 0.8960\n",
      "Epoch 32/100\n",
      "49/49 [==============================] - 534s 11s/step - loss: 0.2530 - accuracy: 0.8998 - val_loss: 0.3525 - val_accuracy: 0.9230\n",
      "Epoch 33/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2557 - accuracy: 0.9016 - val_loss: 0.2519 - val_accuracy: 0.9060\n",
      "Epoch 34/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2507 - accuracy: 0.9055 - val_loss: 0.3765 - val_accuracy: 0.9240\n",
      "Epoch 35/100\n",
      "49/49 [==============================] - 529s 11s/step - loss: 0.2535 - accuracy: 0.9020 - val_loss: 0.2349 - val_accuracy: 0.9120\n",
      "Epoch 36/100\n",
      "49/49 [==============================] - 533s 11s/step - loss: 0.2499 - accuracy: 0.9008 - val_loss: 0.2860 - val_accuracy: 0.9180\n",
      "Epoch 37/100\n",
      "49/49 [==============================] - 533s 11s/step - loss: 0.2509 - accuracy: 0.9026 - val_loss: 0.2721 - val_accuracy: 0.9220\n",
      "Epoch 38/100\n",
      "49/49 [==============================] - 534s 11s/step - loss: 0.2590 - accuracy: 0.8952 - val_loss: 0.2682 - val_accuracy: 0.8970\n",
      "Epoch 39/100\n",
      "49/49 [==============================] - 530s 11s/step - loss: 0.2471 - accuracy: 0.9082 - val_loss: 0.1814 - val_accuracy: 0.9040\n",
      "Epoch 40/100\n",
      "49/49 [==============================] - 532s 11s/step - loss: 0.2510 - accuracy: 0.9028 - val_loss: 0.2524 - val_accuracy: 0.9110\n",
      "Epoch 41/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2430 - accuracy: 0.9065 - val_loss: 0.2250 - val_accuracy: 0.9190\n",
      "Epoch 42/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2484 - accuracy: 0.9041 - val_loss: 0.2073 - val_accuracy: 0.9180\n",
      "Epoch 43/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2495 - accuracy: 0.9020 - val_loss: 0.2198 - val_accuracy: 0.9130\n",
      "Epoch 44/100\n",
      "49/49 [==============================] - 536s 11s/step - loss: 0.2534 - accuracy: 0.9002 - val_loss: 0.3953 - val_accuracy: 0.9000\n",
      "Epoch 45/100\n",
      "49/49 [==============================] - 534s 11s/step - loss: 0.2412 - accuracy: 0.9077 - val_loss: 0.2772 - val_accuracy: 0.9140\n",
      "Epoch 46/100\n",
      "49/49 [==============================] - 530s 11s/step - loss: 0.2519 - accuracy: 0.9038 - val_loss: 0.1476 - val_accuracy: 0.9170\n",
      "Epoch 47/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2412 - accuracy: 0.9059 - val_loss: 0.3050 - val_accuracy: 0.9250\n",
      "Epoch 48/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2423 - accuracy: 0.9054 - val_loss: 0.1610 - val_accuracy: 0.9240\n",
      "Epoch 49/100\n",
      "49/49 [==============================] - 532s 11s/step - loss: 0.2364 - accuracy: 0.9075 - val_loss: 0.1882 - val_accuracy: 0.9230\n",
      "Epoch 50/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2433 - accuracy: 0.9011 - val_loss: 0.2083 - val_accuracy: 0.9200\n",
      "Epoch 51/100\n",
      "49/49 [==============================] - 537s 11s/step - loss: 0.2351 - accuracy: 0.9091 - val_loss: 0.2435 - val_accuracy: 0.9120\n",
      "Epoch 52/100\n",
      "49/49 [==============================] - 532s 11s/step - loss: 0.2330 - accuracy: 0.9071 - val_loss: 0.2243 - val_accuracy: 0.9100\n",
      "Epoch 53/100\n",
      "49/49 [==============================] - 529s 11s/step - loss: 0.2426 - accuracy: 0.9061 - val_loss: 0.2053 - val_accuracy: 0.9210\n",
      "Epoch 54/100\n",
      "49/49 [==============================] - 533s 11s/step - loss: 0.2366 - accuracy: 0.9098 - val_loss: 0.2530 - val_accuracy: 0.9240\n",
      "Epoch 55/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2394 - accuracy: 0.9065 - val_loss: 0.1889 - val_accuracy: 0.9190\n",
      "Epoch 56/100\n",
      "49/49 [==============================] - 532s 11s/step - loss: 0.2342 - accuracy: 0.9054 - val_loss: 0.2240 - val_accuracy: 0.9250\n",
      "Epoch 57/100\n",
      "49/49 [==============================] - 531s 11s/step - loss: 0.2446 - accuracy: 0.9018 - val_loss: 0.2557 - val_accuracy: 0.9050\n",
      "Epoch 58/100\n",
      "49/49 [==============================] - 539s 11s/step - loss: 0.2366 - accuracy: 0.9100 - val_loss: 0.1706 - val_accuracy: 0.9290\n",
      "Epoch 59/100\n",
      "49/49 [==============================] - 530s 11s/step - loss: 0.2369 - accuracy: 0.9069 - val_loss: 0.2842 - val_accuracy: 0.9140\n",
      "Epoch 60/100\n",
      "49/49 [==============================] - 534s 11s/step - loss: 0.2348 - accuracy: 0.9085 - val_loss: 0.2696 - val_accuracy: 0.9190\n",
      "Epoch 61/100\n",
      "49/49 [==============================] - 516s 11s/step - loss: 0.2315 - accuracy: 0.9120 - val_loss: 0.2137 - val_accuracy: 0.9200\n",
      "Epoch 62/100\n",
      "49/49 [==============================] - 506s 10s/step - loss: 0.2342 - accuracy: 0.9071 - val_loss: 0.2431 - val_accuracy: 0.9000\n",
      "Epoch 63/100\n",
      "49/49 [==============================] - 509s 10s/step - loss: 0.2297 - accuracy: 0.9108 - val_loss: 0.2014 - val_accuracy: 0.9170\n",
      "Epoch 64/100\n",
      "49/49 [==============================] - 1031s 21s/step - loss: 0.2386 - accuracy: 0.9016 - val_loss: 0.2333 - val_accuracy: 0.9150\n",
      "Epoch 65/100\n",
      "49/49 [==============================] - 1704s 35s/step - loss: 0.2300 - accuracy: 0.9110 - val_loss: 0.2692 - val_accuracy: 0.9250\n",
      "Epoch 66/100\n",
      "49/49 [==============================] - 418s 9s/step - loss: 0.2353 - accuracy: 0.9096 - val_loss: 0.2395 - val_accuracy: 0.9220\n",
      "Epoch 67/100\n",
      "49/49 [==============================] - 417s 9s/step - loss: 0.2315 - accuracy: 0.9116 - val_loss: 0.2001 - val_accuracy: 0.8820\n",
      "Epoch 68/100\n",
      "49/49 [==============================] - 418s 9s/step - loss: 0.2268 - accuracy: 0.9098 - val_loss: 0.2570 - val_accuracy: 0.8960\n",
      "Epoch 69/100\n",
      "49/49 [==============================] - 435s 9s/step - loss: 0.2221 - accuracy: 0.9116 - val_loss: 0.3449 - val_accuracy: 0.9040\n",
      "Epoch 70/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2313 - accuracy: 0.9086 - val_loss: 0.2989 - val_accuracy: 0.9320\n",
      "Epoch 71/100\n",
      "49/49 [==============================] - 413s 8s/step - loss: 0.2304 - accuracy: 0.9075 - val_loss: 0.3530 - val_accuracy: 0.8980\n",
      "Epoch 72/100\n",
      "49/49 [==============================] - 417s 9s/step - loss: 0.2269 - accuracy: 0.9104 - val_loss: 0.2725 - val_accuracy: 0.9280\n",
      "Epoch 73/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2267 - accuracy: 0.9112 - val_loss: 0.2724 - val_accuracy: 0.9260\n",
      "Epoch 74/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2259 - accuracy: 0.9108 - val_loss: 0.2293 - val_accuracy: 0.9200\n",
      "Epoch 75/100\n",
      "49/49 [==============================] - 414s 8s/step - loss: 0.2279 - accuracy: 0.9106 - val_loss: 0.2245 - val_accuracy: 0.9210\n",
      "Epoch 76/100\n",
      "49/49 [==============================] - 416s 8s/step - loss: 0.2231 - accuracy: 0.9118 - val_loss: 0.2813 - val_accuracy: 0.8980\n",
      "Epoch 77/100\n",
      "49/49 [==============================] - 418s 9s/step - loss: 0.2284 - accuracy: 0.9081 - val_loss: 0.2605 - val_accuracy: 0.9250\n",
      "Epoch 78/100\n",
      "49/49 [==============================] - 417s 9s/step - loss: 0.2212 - accuracy: 0.9147 - val_loss: 0.2709 - val_accuracy: 0.9240\n",
      "Epoch 79/100\n",
      "49/49 [==============================] - 414s 8s/step - loss: 0.2270 - accuracy: 0.9118 - val_loss: 0.2711 - val_accuracy: 0.9020\n",
      "Epoch 80/100\n",
      "49/49 [==============================] - 416s 8s/step - loss: 0.2253 - accuracy: 0.9096 - val_loss: 0.2307 - val_accuracy: 0.9220\n",
      "Epoch 81/100\n",
      "49/49 [==============================] - 414s 8s/step - loss: 0.2218 - accuracy: 0.9146 - val_loss: 0.2019 - val_accuracy: 0.9180\n",
      "Epoch 82/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2298 - accuracy: 0.9075 - val_loss: 0.3207 - val_accuracy: 0.8780\n",
      "Epoch 83/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2243 - accuracy: 0.9096 - val_loss: 0.1574 - val_accuracy: 0.9310\n",
      "Epoch 84/100\n",
      "49/49 [==============================] - 414s 8s/step - loss: 0.2261 - accuracy: 0.9132 - val_loss: 0.2842 - val_accuracy: 0.9240\n",
      "Epoch 85/100\n",
      "49/49 [==============================] - 416s 8s/step - loss: 0.2259 - accuracy: 0.9105 - val_loss: 0.2417 - val_accuracy: 0.9270\n",
      "Epoch 86/100\n",
      "49/49 [==============================] - 420s 9s/step - loss: 0.2281 - accuracy: 0.9122 - val_loss: 0.3169 - val_accuracy: 0.8940\n",
      "Epoch 87/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2166 - accuracy: 0.9147 - val_loss: 0.1545 - val_accuracy: 0.9320\n",
      "Epoch 88/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2303 - accuracy: 0.9085 - val_loss: 0.3089 - val_accuracy: 0.9210\n",
      "Epoch 89/100\n",
      "49/49 [==============================] - 416s 8s/step - loss: 0.2196 - accuracy: 0.9130 - val_loss: 0.2647 - val_accuracy: 0.9270\n",
      "Epoch 90/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2122 - accuracy: 0.9167 - val_loss: 0.1364 - val_accuracy: 0.9210\n",
      "Epoch 91/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2240 - accuracy: 0.9106 - val_loss: 0.1653 - val_accuracy: 0.9000\n",
      "Epoch 92/100\n",
      "49/49 [==============================] - 416s 8s/step - loss: 0.2162 - accuracy: 0.9130 - val_loss: 0.2330 - val_accuracy: 0.9060\n",
      "Epoch 93/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2156 - accuracy: 0.9165 - val_loss: 0.1780 - val_accuracy: 0.9230\n",
      "Epoch 94/100\n",
      "49/49 [==============================] - 417s 9s/step - loss: 0.2139 - accuracy: 0.9183 - val_loss: 0.3739 - val_accuracy: 0.8570\n",
      "Epoch 95/100\n",
      "49/49 [==============================] - 419s 9s/step - loss: 0.2132 - accuracy: 0.9159 - val_loss: 0.2319 - val_accuracy: 0.9260\n",
      "Epoch 96/100\n",
      "49/49 [==============================] - 416s 8s/step - loss: 0.2127 - accuracy: 0.9130 - val_loss: 0.3491 - val_accuracy: 0.9170\n",
      "Epoch 97/100\n",
      "49/49 [==============================] - 414s 8s/step - loss: 0.2160 - accuracy: 0.9158 - val_loss: 0.2129 - val_accuracy: 0.8980\n",
      "Epoch 98/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2191 - accuracy: 0.9140 - val_loss: 0.2627 - val_accuracy: 0.8980\n",
      "Epoch 99/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2142 - accuracy: 0.9165 - val_loss: 0.2314 - val_accuracy: 0.9050\n",
      "Epoch 100/100\n",
      "49/49 [==============================] - 415s 8s/step - loss: 0.2197 - accuracy: 0.9124 - val_loss: 0.2696 - val_accuracy: 0.8610\n"
     ]
    }
   ],
   "source": [
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras.callbacks import ModelCheckpoint, EarlyStopping\n",
    "batch_size = 100\n",
    "epochs = 100\n",
    "\n",
    "early_stopping = EarlyStopping(monitor='val_loss',\n",
    "                              min_delta=0,\n",
    "                              patience=2,\n",
    "                              verbose=0, mode='auto')\n",
    "\n",
    "model.compile(loss='binary_crossentropy',\n",
    "#               optimizer=\"adam\"\n",
    "              optimizer = optimizers.adam(lr=1e-4)\n",
    "              ,\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "# prepare data augmentation configuration\n",
    "train_datagen = ImageDataGenerator(\n",
    "        rescale=1./255,\n",
    "        shear_range=0.2,\n",
    "        zoom_range=0.2,\n",
    "        horizontal_flip=True)\n",
    "\n",
    "\n",
    "test_datagen = ImageDataGenerator(rescale=1./255)\n",
    "\n",
    "train_generator = train_datagen.flow_from_directory(\n",
    "        'data/train',\n",
    "        target_size=(IMG_HEIGHT, IMG_WIDTH),\n",
    "        batch_size=batch_size,\n",
    "        class_mode='binary')\n",
    "\n",
    "validation_generator = test_datagen.flow_from_directory(\n",
    "        'data/validation',\n",
    "        target_size=(IMG_HEIGHT, IMG_WIDTH),\n",
    "        batch_size=batch_size,\n",
    "        class_mode='binary')\n",
    "\n",
    "# fine-tune the model\n",
    "history = model.fit_generator(\n",
    "        train_generator,\n",
    "        steps_per_epoch=4999 // batch_size,\n",
    "        epochs=epochs,\n",
    "        validation_data=validation_generator,\n",
    "        validation_steps= 1000 // batch_size,\n",
    "#         callbacks=[early_stopping]\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "# Plot the loss and accuracy curves for training and validation \n",
    "fig, ax = plt.subplots(2,1)\n",
    "ax[0].plot(history.history['loss'], color='b', label=\"Training loss\")\n",
    "ax[0].plot(history.history['val_loss'], color='r', label=\"validation loss\",axes =ax[0])\n",
    "legend = ax[0].legend(loc='best', shadow=True)\n",
    "\n",
    "ax[1].plot(history.history['accuracy'], color='b', label=\"Training accuracy\")\n",
    "ax[1].plot(history.history['val_accuracy'], color='r',label=\"Validation accuracy\")\n",
    "legend = ax[1].legend(loc='best', shadow=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "scores1 = model.evaluate(train_generator, verbose=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 21.359026432037354\n",
      "Test accuracy: 90.9704864025116\n"
     ]
    }
   ],
   "source": [
    "print ('Test loss:', scores1[0]*100)\n",
    "print ('Test accuracy:', scores1[1]*100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "scores = model.evaluate(validation_generator, verbose=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validation loss: 30.717751383781433\n",
      "Validation accuracy: 86.12499833106995\n"
     ]
    }
   ],
   "source": [
    "print ('Validation loss:', scores[0]*100)\n",
    "print ('Validation accuracy:', scores[1]*100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saved model to disk\n"
     ]
    }
   ],
   "source": [
    "model.save(\"models/vacc8612-vlss3071-tacc9097-tlss2135.h5\")\n",
    "print(\"Saved model to disk\")"
   ]
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}