deeplearning_project_vgg16.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:3976: The name tf.nn.max_pool is deprecated. Please use tf.nn.max_pool2d instead.\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:174: The name tf.get_default_session is deprecated. Please use tf.compat.v1.get_default_session instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:181: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:186: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:190: The name tf.global_variables is deprecated. Please use tf.compat.v1.global_variables instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:199: The name tf.is_variable_initialized is deprecated. Please use tf.compat.v1.is_variable_initialized instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:206: The name tf.variables_initializer is deprecated. Please use tf.compat.v1.variables_initializer instead.\n",
      "\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 224, 224, 3)       0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, 224, 224, 64)      1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, 224, 224, 64)      36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, 112, 112, 64)      0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, 112, 112, 128)     73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, 112, 112, 128)     147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, 56, 56, 128)       0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, 56, 56, 256)       295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, 28, 28, 256)       0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, 28, 28, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, 14, 14, 512)       0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, 7, 7, 512)         0         \n",
      "=================================================================\n",
      "Total params: 14,714,688\n",
      "Trainable params: 14,714,688\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Found 12000 images belonging to 50 classes.\n",
      "Found 3000 images belonging to 50 classes.\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:133: The name tf.placeholder_with_default is deprecated. Please use tf.compat.v1.placeholder_with_default instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "max_pooling2d_1 (MaxPooling2 (None, 3, 3, 512)         0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 4608)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 4096)              18878464  \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 4096)              16781312  \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 1000)              4097000   \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 1000)              0         \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 1000)              0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 50)                50050     \n",
      "_________________________________________________________________\n",
      "activation_4 (Activation)    (None, 50)                0         \n",
      "=================================================================\n",
      "Total params: 39,806,826\n",
      "Trainable params: 39,806,826\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\optimizers.py:790: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:3295: The name tf.log is deprecated. Please use tf.math.log instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\tensorflow_core\\python\\ops\\math_grad.py:1424: where (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.where in 2.0, which has the same broadcast rule as np.where\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:986: The name tf.assign_add is deprecated. Please use tf.compat.v1.assign_add instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\maste\\.conda\\envs\\myenv\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:973: The name tf.assign is deprecated. Please use tf.compat.v1.assign instead.\n",
      "\n",
      "Train on 12000 samples, validate on 3000 samples\n",
      "Epoch 1/200\n",
      "12000/12000 [==============================] - 33s 3ms/step - loss: 2.1786 - acc: 0.3655 - val_loss: 1.2888 - val_acc: 0.6010\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 1.28878, saving model to vgg16_mod.h5\n",
      "Epoch 2/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "12000/12000 [==============================] - 31s 3ms/step - loss: 1.3054 - acc: 0.5827 - val_loss: 0.9087 - val_acc: 0.7267\n",
      "\n",
      "Epoch 00002: val_loss improved from 1.28878 to 0.90873, saving model to vgg16_mod.h5\n",
      "Epoch 3/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 1.2331 - acc: 0.6094 - val_loss: 0.8867 - val_acc: 0.7103\n",
      "\n",
      "Epoch 00003: val_loss improved from 0.90873 to 0.88669, saving model to vgg16_mod.h5\n",
      "Epoch 4/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 1.0704 - acc: 0.6555 - val_loss: 0.7555 - val_acc: 0.7593\n",
      "\n",
      "Epoch 00004: val_loss improved from 0.88669 to 0.75549, saving model to vgg16_mod.h5\n",
      "Epoch 5/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.9256 - acc: 0.6988 - val_loss: 0.8332 - val_acc: 0.7443\n",
      "\n",
      "Epoch 00005: val_loss did not improve from 0.75549\n",
      "Epoch 6/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.8525 - acc: 0.7230 - val_loss: 0.5811 - val_acc: 0.8130\n",
      "\n",
      "Epoch 00006: val_loss improved from 0.75549 to 0.58111, saving model to vgg16_mod.h5\n",
      "Epoch 7/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.7620 - acc: 0.7518 - val_loss: 0.5853 - val_acc: 0.8223\n",
      "\n",
      "Epoch 00007: val_loss did not improve from 0.58111\n",
      "Epoch 8/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.6766 - acc: 0.7827 - val_loss: 0.5154 - val_acc: 0.8360\n",
      "\n",
      "Epoch 00008: val_loss improved from 0.58111 to 0.51544, saving model to vgg16_mod.h5\n",
      "Epoch 9/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.6401 - acc: 0.7907 - val_loss: 0.5319 - val_acc: 0.8420\n",
      "\n",
      "Epoch 00009: val_loss did not improve from 0.51544\n",
      "Epoch 10/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.6119 - acc: 0.8038 - val_loss: 0.6082 - val_acc: 0.8070\n",
      "\n",
      "Epoch 00010: val_loss did not improve from 0.51544\n",
      "Epoch 11/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.5915 - acc: 0.8106 - val_loss: 0.5604 - val_acc: 0.8283\n",
      "\n",
      "Epoch 00011: val_loss did not improve from 0.51544\n",
      "Epoch 12/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.5243 - acc: 0.8261 - val_loss: 0.4370 - val_acc: 0.8597\n",
      "\n",
      "Epoch 00012: val_loss improved from 0.51544 to 0.43704, saving model to vgg16_mod.h5\n",
      "Epoch 13/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.4737 - acc: 0.8454 - val_loss: 0.5336 - val_acc: 0.8407\n",
      "\n",
      "Epoch 00013: val_loss did not improve from 0.43704\n",
      "Epoch 14/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.4795 - acc: 0.8467 - val_loss: 0.4312 - val_acc: 0.8750\n",
      "\n",
      "Epoch 00014: val_loss improved from 0.43704 to 0.43116, saving model to vgg16_mod.h5\n",
      "Epoch 15/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.4213 - acc: 0.8607 - val_loss: 0.5093 - val_acc: 0.8323\n",
      "\n",
      "Epoch 00015: val_loss did not improve from 0.43116\n",
      "Epoch 16/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.4178 - acc: 0.8679 - val_loss: 0.4626 - val_acc: 0.8613\n",
      "\n",
      "Epoch 00016: val_loss did not improve from 0.43116\n",
      "Epoch 17/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.3957 - acc: 0.8708 - val_loss: 0.4305 - val_acc: 0.8703\n",
      "\n",
      "Epoch 00017: val_loss improved from 0.43116 to 0.43048, saving model to vgg16_mod.h5\n",
      "Epoch 18/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.3821 - acc: 0.8758 - val_loss: 0.4510 - val_acc: 0.8657\n",
      "\n",
      "Epoch 00018: val_loss did not improve from 0.43048\n",
      "Epoch 19/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.3677 - acc: 0.8797 - val_loss: 0.3966 - val_acc: 0.8763\n",
      "\n",
      "Epoch 00019: val_loss improved from 0.43048 to 0.39656, saving model to vgg16_mod.h5\n",
      "Epoch 20/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.3575 - acc: 0.8826 - val_loss: 0.4489 - val_acc: 0.8603\n",
      "\n",
      "Epoch 00020: val_loss did not improve from 0.39656\n",
      "Epoch 21/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.3627 - acc: 0.8825 - val_loss: 0.4934 - val_acc: 0.8467\n",
      "\n",
      "Epoch 00021: val_loss did not improve from 0.39656\n",
      "Epoch 22/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.3873 - acc: 0.8753 - val_loss: 0.3838 - val_acc: 0.8797\n",
      "\n",
      "Epoch 00022: val_loss improved from 0.39656 to 0.38377, saving model to vgg16_mod.h5\n",
      "Epoch 23/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.3405 - acc: 0.8928 - val_loss: 0.4038 - val_acc: 0.8760\n",
      "\n",
      "Epoch 00023: val_loss did not improve from 0.38377\n",
      "Epoch 24/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.3358 - acc: 0.8911 - val_loss: 0.4242 - val_acc: 0.8707\n",
      "\n",
      "Epoch 00024: val_loss did not improve from 0.38377\n",
      "Epoch 25/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.3331 - acc: 0.8899 - val_loss: 0.4428 - val_acc: 0.8773\n",
      "\n",
      "Epoch 00025: val_loss did not improve from 0.38377\n",
      "Epoch 26/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.3150 - acc: 0.8957 - val_loss: 0.4268 - val_acc: 0.8730\n",
      "\n",
      "Epoch 00026: val_loss did not improve from 0.38377\n",
      "Epoch 27/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.2789 - acc: 0.9078 - val_loss: 0.3845 - val_acc: 0.8903\n",
      "\n",
      "Epoch 00027: val_loss did not improve from 0.38377\n",
      "Epoch 28/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.1374 - acc: 0.9558 - val_loss: 0.2490 - val_acc: 0.9300\n",
      "\n",
      "Epoch 00028: val_loss improved from 0.38377 to 0.24901, saving model to vgg16_mod.h5\n",
      "Epoch 29/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.1011 - acc: 0.9657 - val_loss: 0.2504 - val_acc: 0.9300\n",
      "\n",
      "Epoch 00029: val_loss did not improve from 0.24901\n",
      "Epoch 30/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0892 - acc: 0.9713 - val_loss: 0.2456 - val_acc: 0.9303\n",
      "\n",
      "Epoch 00030: val_loss improved from 0.24901 to 0.24557, saving model to vgg16_mod.h5\n",
      "Epoch 31/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0919 - acc: 0.9696 - val_loss: 0.2559 - val_acc: 0.9290\n",
      "\n",
      "Epoch 00031: val_loss did not improve from 0.24557\n",
      "Epoch 32/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0778 - acc: 0.9736 - val_loss: 0.2453 - val_acc: 0.9323\n",
      "\n",
      "Epoch 00032: val_loss improved from 0.24557 to 0.24531, saving model to vgg16_mod.h5\n",
      "Epoch 33/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0785 - acc: 0.9727 - val_loss: 0.2442 - val_acc: 0.9327\n",
      "\n",
      "Epoch 00033: val_loss improved from 0.24531 to 0.24420, saving model to vgg16_mod.h5\n",
      "Epoch 34/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0781 - acc: 0.9744 - val_loss: 0.2500 - val_acc: 0.9330\n",
      "\n",
      "Epoch 00034: val_loss did not improve from 0.24420\n",
      "Epoch 35/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0767 - acc: 0.9738 - val_loss: 0.2473 - val_acc: 0.9313\n",
      "\n",
      "Epoch 00035: val_loss did not improve from 0.24420\n",
      "Epoch 36/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0691 - acc: 0.9767 - val_loss: 0.2421 - val_acc: 0.9340\n",
      "\n",
      "Epoch 00036: val_loss improved from 0.24420 to 0.24211, saving model to vgg16_mod.h5\n",
      "Epoch 37/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0712 - acc: 0.9762 - val_loss: 0.2542 - val_acc: 0.9317\n",
      "\n",
      "Epoch 00037: val_loss did not improve from 0.24211\n",
      "Epoch 38/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0724 - acc: 0.9761 - val_loss: 0.2507 - val_acc: 0.9307c: 0.9\n",
      "\n",
      "Epoch 00038: val_loss did not improve from 0.24211\n",
      "Epoch 39/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0655 - acc: 0.9783 - val_loss: 0.2514 - val_acc: 0.9310\n",
      "\n",
      "Epoch 00039: val_loss did not improve from 0.24211\n",
      "Epoch 40/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0686 - acc: 0.9767 - val_loss: 0.2471 - val_acc: 0.9333\n",
      "\n",
      "Epoch 00040: val_loss did not improve from 0.24211\n",
      "Epoch 41/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0680 - acc: 0.9780 - val_loss: 0.2649 - val_acc: 0.9297\n",
      "\n",
      "Epoch 00041: val_loss did not improve from 0.24211\n",
      "Epoch 42/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0612 - acc: 0.9787 - val_loss: 0.2520 - val_acc: 0.9300\n",
      "\n",
      "Epoch 00042: val_loss did not improve from 0.24211\n",
      "Epoch 43/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0583 - acc: 0.9799 - val_loss: 0.2482 - val_acc: 0.9340\n",
      "\n",
      "Epoch 00043: val_loss did not improve from 0.24211\n",
      "Epoch 44/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0576 - acc: 0.9818 - val_loss: 0.2485 - val_acc: 0.9337\n",
      "\n",
      "Epoch 00044: val_loss did not improve from 0.24211\n",
      "Epoch 45/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0600 - acc: 0.9808 - val_loss: 0.2480 - val_acc: 0.9330\n",
      "\n",
      "Epoch 00045: val_loss did not improve from 0.24211\n",
      "Epoch 46/200\n",
      "12000/12000 [==============================] - 31s 3ms/step - loss: 0.0536 - acc: 0.9822 - val_loss: 0.2478 - val_acc: 0.9333\n",
      "\n",
      "Epoch 00046: val_loss did not improve from 0.24211\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Convolution2D, MaxPooling2D, Flatten, Dense, Dropout, GlobalAveragePooling2D,Activation\n",
    "from keras.applications import VGG16\n",
    "\n",
    "train_dir = \"C://Users//maste//Desktop//dl_project//data1//BasicFinalDatabase//Train\"\n",
    "valid_dir = \"C://Users//maste//Desktop//dl_project//data1//BasicFinalDatabase/Test\"\n",
    "\n",
    "img_width, img_height = 224, 224\n",
    "\n",
    "conv_base = VGG16(weights='imagenet', \n",
    "                  include_top=False,\n",
    "                  input_shape=(img_width, img_height, 3))\n",
    "\n",
    "# Show architecture\n",
    "conv_base.summary()\n",
    "\n",
    "\n",
    "import os, shutil\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "datagen = ImageDataGenerator(rescale=1./255)\n",
    "batch_size = 24\n",
    "\n",
    "def extract_features(directory, sample_count):\n",
    "    features = np.zeros(shape=(sample_count, 7, 7, 512))  # Must be equal to the output of the convolutional base\n",
    "    labels = np.zeros(shape=(sample_count,50))\n",
    "    # Preprocess data\n",
    "    generator = datagen.flow_from_directory(directory,\n",
    "                                            target_size=(img_width,img_height),\n",
    "                                            batch_size = batch_size,\n",
    "                                            class_mode='categorical')\n",
    "    # Pass data through convolutional base\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\n",
    "        labels[i * batch_size: (i + 1) * batch_size] = labels_batch\n",
    "        i += 1\n",
    "        if i * batch_size >= sample_count:\n",
    "            break\n",
    "    return features, labels\n",
    "\n",
    "\n",
    "train_features, train_labels = extract_features(train_dir, 240*50)\n",
    "validation_features, validation_labels = extract_features(valid_dir, 60*50)\n",
    "\n",
    "epochs = 200\n",
    "\n",
    "model = Sequential()\n",
    "\n",
    "\n",
    "model.add(MaxPooling2D(input_shape=(7,7,512),pool_size=(2,2), strides=(2,2), padding='valid'))\n",
    "\n",
    "\n",
    "model.add(Flatten())\n",
    "\n",
    "model.add(Dense(4096))\n",
    "model.add(Activation('relu'))\n",
    "\n",
    "model.add(Dropout(0.15))\n",
    "\n",
    "\n",
    "model.add(Dense(4096))\n",
    "model.add(Activation('relu'))\n",
    "\n",
    "model.add(Dropout(0.15))\n",
    "\n",
    "\n",
    "model.add(Dense(1000))\n",
    "model.add(Activation('relu'))\n",
    "\n",
    "model.add(Dropout(0.15))\n",
    "\n",
    "model.add(Dense(50))\n",
    "model.add(Activation('softmax'))\n",
    "model.summary()\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "from keras.callbacks import ModelCheckpoint,ReduceLROnPlateau,EarlyStopping\n",
    "checkpoint = ModelCheckpoint('vgg16_mod.h5', verbose=1, monitor='val_loss',save_best_only=True, mode='auto')  \n",
    "red=ReduceLROnPlateau(monitor=\"val_loss\",factor=0.1,patience=5,mode=\"auto\")\n",
    "early=EarlyStopping(monitor=\"val_loss\",min_delta=1e-4,patience=10,mode=\"auto\")\n",
    "# Compile model\n",
    "from keras.optimizers import Adam,SGD\n",
    "sgd=SGD(lr=0.01,decay=1e-6,momentum=0.9,nesterov=True)\n",
    "model.compile(optimizer=sgd,\n",
    "              loss='categorical_crossentropy',\n",
    "              metrics=['acc'])\n",
    "\n",
    "# Train model\n",
    "history = model.fit(train_features, train_labels,\n",
    "                    epochs=epochs,\n",
    "                    batch_size=batch_size, \n",
    "                    callbacks=[checkpoint,red,early],\n",
    "                    validation_data=(validation_features, validation_labels))\n",
    "\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "\"\\nplt.plot(epochs, loss, 'g', label='Training loss')\\nplt.plot(epochs, val_loss, 'r', label='Validation loss')\\nplt.title('Training and validation loss')\\nplt.legend()\\n\\nplt.show()\\n\""
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "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, 'g', label='Training accuracy')\n",
    "plt.plot(epochs, val_acc, 'r', label='Validation accuracy')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "plt.show()\n",
    "'''\n",
    "plt.plot(epochs, loss, 'g', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'r', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()\n",
    "'''\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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"
    }
   ],
   "source": [
    "plt.plot(epochs, loss, 'g', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'r', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "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": "3.7.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}