code/initial-model-comparison/nasnetmobile_finetune_only_fc.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "1d9539215c52a8206c18eaebb0d6045087c76d74"
   },
   "source": [
    "# **FINETUNING NASNetMobile MODEL FOR DETERMINING RSC**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "a3bae869c049172a16bd52f66438861590b9006a"
   },
   "source": [
    "## **Import libraries**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5"
   },
   "outputs": [],
   "source": [
    "# Libraries for file and data handling\n",
    "%matplotlib inline\n",
    "import os, sys\n",
    "import numpy as np \n",
    "import pandas as pd\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "from scipy import misc\n",
    "import imageio\n",
    "from IPython.display import display\n",
    "from IPython.display import Image as _Imgdis\n",
    "from PIL import Image\n",
    "import pickle\n",
    "from random import shuffle, seed\n",
    "seed(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "79c7e3d0-c299-4dcb-8224-4455121ee9b0",
    "_uuid": "d629ff2d2480ee46fbb7e2d37f6b5fab8052498a"
   },
   "source": [
    "## **Data preparation**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load the lists detailing the train and test files\n",
    "with open('../input/rwis-data-split/split.pickle', 'rb') as handle:\n",
    "    split_dic = pickle.load(handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Images in train set:12600\n",
      "Images in test set:1400\n"
     ]
    }
   ],
   "source": [
    "# Obtain the lists\n",
    "train = split_dic['train']\n",
    "print('Images in train set:' + str(len(train)))\n",
    "test = split_dic['test']\n",
    "print('Images in test set:' + str(len(test)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Image dimensions\n",
    "image_height = 224\n",
    "image_width = 224\n",
    "channels = 3\n",
    "display_images = True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "label_names = ['Bare pavement', 'Partial snow coverage', 'Full snow coverage']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Function that prints the total number of images per category in the given set\n",
    "def print_items(list_of_files):\n",
    "    cat1 = 0\n",
    "    cat2 = 0\n",
    "    cat3 = 0\n",
    "    for ifile in list_of_files:\n",
    "        if 'C1' in ifile:\n",
    "            cat1 += 1\n",
    "        elif 'C2' in ifile:\n",
    "            cat2 += 1\n",
    "        elif 'C3' in ifile:\n",
    "            cat3 += 1\n",
    "    total = cat1+cat2+cat3\n",
    "    print ('Total images: ' + str(total))\n",
    "    print (label_names[0] + ' ' + str(cat1) + ' Approx. ' + str(int(cat1/total*100)) + '%')\n",
    "    print (label_names[1] + ' ' + str(cat2) + ' Approx. ' + str(int(cat2/total*100)) + '%')\n",
    "    print (label_names[2] + ' ' + str(cat3) + ' Approx. ' + str(int(cat3/total*100)) + '%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "- Summary of images in the train set -\n",
      "Total images: 12600\n",
      "Bare pavement 5691 Approx. 45%\n",
      "Partial snow coverage 5114 Approx. 40%\n",
      "Full snow coverage 1795 Approx. 14%\n"
     ]
    }
   ],
   "source": [
    "print('- Summary of images in the train set -')\n",
    "print_items(train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "- Summary of images in the test set -\n",
      "Total images: 1400\n",
      "Bare pavement 648 Approx. 46%\n",
      "Partial snow coverage 558 Approx. 39%\n",
      "Full snow coverage 194 Approx. 13%\n"
     ]
    }
   ],
   "source": [
    "print('- Summary of images in the test set -')\n",
    "print_items(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Function that outputs the list of all files in the directory tree \n",
    "def getListOfFiles(dirName):\n",
    "    # create a list of file and sub directories \n",
    "    # names in the given directory \n",
    "    listOfFile = os.listdir(dirName)\n",
    "    allFiles = list()\n",
    "    # Iterate over all the entries\n",
    "    for entry in listOfFile:\n",
    "        # Create full path\n",
    "        fullPath = os.path.join(dirName, entry)\n",
    "        # If entry is a directory then get the list of files in this directory \n",
    "        if os.path.isdir(fullPath):\n",
    "            allFiles = allFiles + getListOfFiles(fullPath)\n",
    "        else:\n",
    "            if '.jpg' in fullPath :\n",
    "                allFiles.append(fullPath)    \n",
    "    return allFiles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Function to use the list of image filenames and import images into numpy X and Y arrays\n",
    "def getArrays(folder, list_of_files):\n",
    "    \n",
    "    # Number of images in the list\n",
    "    numim = len(list_of_files)\n",
    "    # Prepare array for the images\n",
    "    images_array = np.ndarray(shape=(numim, image_height, image_width, channels), dtype=np.float16)\n",
    "    # Prepare array for the labels\n",
    "    labels_array = np.ndarray(shape = numim,dtype=np.uint8)\n",
    "    # Set some counters\n",
    "    idx_files = 0\n",
    "    completion = 20\n",
    "    \n",
    "    # Loop over the filenames\n",
    "    for ifile in list_of_files:\n",
    "        fullpath = folder + ifile\n",
    "        # From filename to numpy array\n",
    "        img_arr = imageio.imread(fullpath)\n",
    "        images_array[idx_files] = img_arr / 255.0\n",
    "        # Obtain label from filename\n",
    "        if '/C1/' in ifile:\n",
    "            label = 0\n",
    "        elif '/C2/' in ifile:\n",
    "            label = 1\n",
    "        elif '/C3/' in ifile:\n",
    "            label = 2\n",
    "        labels_array[idx_files] = label\n",
    "        idx_files += 1\n",
    "        \n",
    "        # Print % of completion\n",
    "        if (idx_files+1) % (numim//5) == 0:\n",
    "            print('Completed ' + str(completion) + '%')\n",
    "            completion += 20\n",
    "\n",
    "    return images_array, labels_array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "folder = \"../input/rwis224/rwis224/rwis224\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Completed 20%\n",
      "Completed 40%\n",
      "Completed 60%\n",
      "Completed 80%\n",
      "Completed 100%\n",
      "X_train shape is (12600, 224, 224, 3)\n",
      "y_train shape is (12600,)\n"
     ]
    }
   ],
   "source": [
    "# Import the corresponding images into numpy arrays \n",
    "X_train, y_train = getArrays(folder, train)\n",
    "print('X_train shape is ' + str(X_train.shape))\n",
    "print('y_train shape is ' + str(y_train.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Completed 20%\n",
      "Completed 40%\n",
      "Completed 60%\n",
      "Completed 80%\n",
      "Completed 100%\n",
      "X_test shape is (1400, 224, 224, 3)\n",
      "y_test shape is (1400,)\n"
     ]
    }
   ],
   "source": [
    "# Import the corresponding images into numpy arrays \n",
    "X_test, y_test = getArrays(folder, test)\n",
    "print('X_test shape is ' + str(X_test.shape))\n",
    "print('y_test shape is ' + str(y_test.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/ER-14/C2/ER-Hwy401_Curry_Hill-p3-1515260700.jpg\n"
     ]
    },
    {
     "data": {
      "image/jpeg": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {
      "image/jpeg": {
       "height": 224,
       "width": 224
      }
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/ER-32/C3/ER-Hwy416_Rideau_River-p3-1515841200.jpg\n"
     ]
    },
    {
     "data": {
      "image/jpeg": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {
      "image/jpeg": {
       "height": 224,
       "width": 224
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Display some images from the train set\n",
    "if display_images:    \n",
    "    for i in range(0, 2):\n",
    "        print(train[i])\n",
    "        display(_Imgdis(filename=folder+train[i], width = image_width, height = image_height))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/CR-19/C2/CR-Hwy401_near_Newtonville_Rd-p1-1513948500.jpg\n"
     ]
    },
    {
     "data": {
      "image/jpeg": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {
      "image/jpeg": {
       "height": 224,
       "width": 224
      }
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/NER-14/C1/NER-Hwy11_Marten_River-p1-1515527100.jpg\n"
     ]
    },
    {
     "data": {
      "image/jpeg": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {
      "image/jpeg": {
       "height": 224,
       "width": 224
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Display some images from the test set\n",
    "if display_images:\n",
    "    for i in range(0, 2):\n",
    "        print(test[i])\n",
    "        display(_Imgdis(filename=folder+test[i], width = image_width, height = image_height))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "fc0783ecde5a919142cf4fe58aa99069fe1ee6c5"
   },
   "source": [
    "## **Model preparation**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "_uuid": "0625cf058cd9028498ca6145bb6b97292ae4e386"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "#Import model specific libraries\n",
    "from keras.applications.nasnet import NASNetMobile\n",
    "from keras.preprocessing import image\n",
    "from keras.models import Model\n",
    "from keras.layers import Dense, GlobalAveragePooling2D\n",
    "from keras import backend as K"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "_uuid": "02b9c493c03a60b67f95e61a25e2f530ecee9607"
   },
   "outputs": [],
   "source": [
    "# Import other Keras libraries\n",
    "from time import time\n",
    "from keras.utils import np_utils\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout, Flatten, Input\n",
    "from keras.constraints import maxnorm\n",
    "from keras import optimizers\n",
    "from keras.utils import to_categorical\n",
    "from keras.layers.convolutional import Conv2D, MaxPooling2D\n",
    "from keras.callbacks import TensorBoard"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "_uuid": "9774c3a243fd80ebbf9b9f7b4bdc2363e3ef774f"
   },
   "outputs": [],
   "source": [
    "# fix random seed for reproducibility\n",
    "seed = 7\n",
    "np.random.seed(seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "_uuid": "ac3fb849084961bb762a3aca87d9d299b834a319"
   },
   "outputs": [],
   "source": [
    "# Convert class vectors to binary class matrices.\n",
    "num_classes = 3\n",
    "y_train = to_categorical(y_train, num_classes)\n",
    "y_test = to_categorical(y_test, num_classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "_uuid": "5e890731b227abf09f640194b84c8bd8e094538a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(12600, 3)\n",
      "(1400, 3)\n"
     ]
    }
   ],
   "source": [
    "print(y_train.shape)\n",
    "print(y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "1d40bb5f260c834e62b34f545b07c634ec4cdebd"
   },
   "source": [
    "**Defining the CNN architecture**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "_uuid": "f5e906a3795d47dfe497565112c5dbaaadf84881"
   },
   "outputs": [],
   "source": [
    "#  Import the base model for transfer learning\n",
    "weights_source = '../input/nasnetmobile-weights/NASNet-mobile-no-top.h5'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "_uuid": "69a8d3bc79cb17f6641048b040813c6c11b5420d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /opt/conda/lib/python3.6/site-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n"
     ]
    }
   ],
   "source": [
    "# create the base pre-trained model\n",
    "base_model = NASNetMobile(weights=weights_source, include_top=False, input_shape=(image_height,image_width,channels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "_uuid": "f8adb9a96c69ee7d27c679f771be62035e41f6da"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /opt/conda/lib/python3.6/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"
     ]
    }
   ],
   "source": [
    "# Use output from base model\n",
    "x = base_model.output\n",
    "# In the original NasNetMobile architecture there is a GAP2D layer before the FC layers\n",
    "x = GlobalAveragePooling2D()(x)\n",
    "# Use same \"head\" or classifier as the baseline model\n",
    "x = Dropout(0.5)(x)\n",
    "# No need for Flatten layer\n",
    "# x = Flatten()(x)\n",
    "x = Dense(48, activation='relu')(x)\n",
    "x = Dropout(0.5)(x)\n",
    "x = Dense(24, activation='relu')(x)\n",
    "x = Dropout(0.5)(x)\n",
    "predictions = Dense(3, activation='softmax')(x)\n",
    "\n",
    "# this is the model we will train\n",
    "model = Model(inputs=base_model.input, outputs=predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "view_model = False\n",
    "if view_model:\n",
    "    print(model.summary())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "c7fbcf10e60c48fb76767c63cdd54817e6c1d1a4"
   },
   "source": [
    "## **Model training**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "_uuid": "3de91cb2c51d5c9427e9599bea6455b6202879b3"
   },
   "outputs": [],
   "source": [
    "# In this case only train the \"head\", or classifier\n",
    "# Set all layers in the base model as not trainable\n",
    "for layer in base_model.layers:\n",
    "    layer.trainable = False\n",
    "\n",
    "# compile the model (should be done *after* setting layers to non-trainable)\n",
    "opt = optimizers.SGD(lr=0.001, momentum=0.9, nesterov=True)\n",
    "model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "_uuid": "6a9709e9b473a784320d14466991c08933f262e7"
   },
   "outputs": [],
   "source": [
    "batch_size = 32\n",
    "epochs = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "_uuid": "7412eaa4056ce1decdb87d3e35b69e7574795fac"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /opt/conda/lib/python3.6/site-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n",
      "Train on 10080 samples, validate on 2520 samples\n",
      "Epoch 1/50\n",
      "10080/10080 [==============================] - 46s 5ms/step - loss: 1.0901 - acc: 0.4197 - val_loss: 0.9886 - val_acc: 0.4563\n",
      "Epoch 2/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 1.0028 - acc: 0.4646 - val_loss: 0.9843 - val_acc: 0.4595\n",
      "Epoch 3/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.9872 - acc: 0.4592 - val_loss: 0.9741 - val_acc: 0.4734\n",
      "Epoch 4/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.9733 - acc: 0.4752 - val_loss: 0.9689 - val_acc: 0.4750\n",
      "Epoch 5/50\n",
      "10080/10080 [==============================] - 31s 3ms/step - loss: 0.9661 - acc: 0.4798 - val_loss: 0.9610 - val_acc: 0.4952\n",
      "Epoch 6/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.9532 - acc: 0.4988 - val_loss: 0.9504 - val_acc: 0.5095\n",
      "Epoch 7/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.9460 - acc: 0.5095 - val_loss: 0.9444 - val_acc: 0.5210\n",
      "Epoch 8/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.9368 - acc: 0.5199 - val_loss: 0.9317 - val_acc: 0.5425\n",
      "Epoch 9/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.9250 - acc: 0.5251 - val_loss: 0.9248 - val_acc: 0.5718\n",
      "Epoch 10/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.9168 - acc: 0.5341 - val_loss: 0.9254 - val_acc: 0.5619\n",
      "Epoch 11/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.9028 - acc: 0.5472 - val_loss: 0.9118 - val_acc: 0.5683\n",
      "Epoch 12/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.9000 - acc: 0.5534 - val_loss: 0.9098 - val_acc: 0.5643\n",
      "Epoch 13/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.8890 - acc: 0.5586 - val_loss: 0.8978 - val_acc: 0.5714\n",
      "Epoch 14/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.8779 - acc: 0.5679 - val_loss: 0.8988 - val_acc: 0.5683\n",
      "Epoch 15/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.8756 - acc: 0.5698 - val_loss: 0.8925 - val_acc: 0.5690\n",
      "Epoch 16/50\n",
      "10080/10080 [==============================] - 31s 3ms/step - loss: 0.8737 - acc: 0.5805 - val_loss: 0.8966 - val_acc: 0.5556\n",
      "Epoch 17/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.8617 - acc: 0.5851 - val_loss: 0.8882 - val_acc: 0.5659\n",
      "Epoch 18/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.8557 - acc: 0.5862 - val_loss: 0.8829 - val_acc: 0.5567\n",
      "Epoch 19/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.8531 - acc: 0.5917 - val_loss: 0.8701 - val_acc: 0.5833\n",
      "Epoch 20/50\n",
      "10080/10080 [==============================] - 30s 3ms/step - loss: 0.8488 - acc: 0.5962 - val_loss: 0.8706 - val_acc: 0.5742\n",
      "Epoch 21/50\n",
      " 2816/10080 [=======>......................] - ETA: 17s - loss: 0.8331 - acc: 0.6072"
     ]
    }
   ],
   "source": [
    "train_history = model.fit(x=X_train, y=y_train, batch_size=batch_size, \n",
    "                      epochs=epochs, verbose=1,\n",
    "                      validation_split = 0.2,\n",
    "                      shuffle=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "1d81de93d2d68d345dd6cae77f1a4d42bf2e71e4"
   },
   "source": [
    "## **Evaluate results**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "_uuid": "d55dedb470f835ed7035e4322d659ddfd12a8ebc"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Get the values from the training history dictionary\n",
    "train_loss = train_history.history['loss'] \n",
    "val_loss = train_history.history['val_loss']\n",
    "\n",
    "# Plot the resulting loss\n",
    "plt.plot(range(1,1+len(train_loss)), train_loss)\n",
    "plt.plot(range(1,1+len(val_loss)), val_loss)\n",
    "plt.xticks(np.arange(1,1+len(train_loss),5))\n",
    "plt.title('Model loss')\n",
    "plt.ylabel('Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Validation'], loc='upper left')\n",
    "axes = plt.gca()\n",
    "axes.set_xlim([1,len(train_loss)])\n",
    "axes.set_ylim([0,5])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "_uuid": "fce403719398d79f1b7b4ae6bbac23425684954e"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Get the values from the training history dictionary\n",
    "train_acc = train_history.history['acc'] \n",
    "val_acc = train_history.history['val_acc']\n",
    "\n",
    "# Plot the resulting loss\n",
    "plt.plot(range(1,1+len(train_acc)), train_acc)\n",
    "plt.plot(range(1,1+len(val_acc)), val_acc)\n",
    "plt.xticks(np.arange(1,1+len(train_acc),5))\n",
    "plt.title('Model accuracy')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Validation'], loc='upper right')\n",
    "axes = plt.gca()\n",
    "axes.set_xlim([1,len(train_acc)])\n",
    "axes.set_ylim([0,1])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--Evaluation over the train and validation datasets--\n",
      "Final train loss: 0.7652\n",
      "Final validation loss: 0.8840\n",
      "Final train accuracy:  65.53%\n",
      "Final validation accuracy:  55.83%\n"
     ]
    }
   ],
   "source": [
    "print(\"--Evaluation over the train and validation datasets--\")\n",
    "print(\"Final train loss: \"+'{:6.4f}'.format(train_loss[-1]))   \n",
    "print(\"Final validation loss: \"+'{:6.4f}'.format(val_loss[-1]))   \n",
    "print(\"Final train accuracy: \"+'{:6.2f}'.format(train_acc[-1]*100)+'%')\n",
    "print(\"Final validation accuracy: \"+'{:6.2f}'.format(val_acc[-1]*100)+'%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "_uuid": "ffeb167fbd2d68ce727ef8fea4a10d3b012ee999"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--Evaluation over the test dataset--\n",
      "Loss: 0.8881\n",
      "Accuracy:  55.64%\n"
     ]
    }
   ],
   "source": [
    "scores = model.evaluate(X_test, y_test, verbose=0)\n",
    "print(\"--Evaluation over the test dataset--\")\n",
    "print(\"Loss: \"+'{:6.4f}'.format(scores[0]))   \n",
    "print(\"Accuracy: \"+'{:6.2f}'.format(scores[1]*100)+'%')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_uuid": "8e1409de9f31a89246dc65c90c2b2c1958e2f975"
   },
   "source": [
    "### Save results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "_uuid": "b26ca9da7ba2470026f28595d91ab559cb1cd49d"
   },
   "outputs": [],
   "source": [
    "nasnetmobile_history_dic = {'train_loss':train_loss,\n",
    "                    'val_loss':val_loss,\n",
    "                     'train_acc':train_acc,\n",
    "                      'val_acc':val_acc}\n",
    "\n",
    "with open('nasnetmobile_history_only_head.pickle', 'wb') as handle:\n",
    "    pickle.dump(nasnetmobile_history_dic, handle, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}