face_shape.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "start_time = time.time()\n",
    "import os\n",
    "import numpy as np\n",
    "from keras import optimizers\n",
    "from keras.preprocessing import image\n",
    "from keras.layers import Dense, Conv2D, MaxPooling2D, BatchNormalization, GlobalAveragePooling2D\n",
    "from keras.layers import Activation, Dropout, Flatten, Dense\n",
    "from keras.models import Sequential\n",
    "from keras import backend as K\n",
    "from keras.optimizers import adam\n",
    "from keras import models\n",
    "import cv2\n",
    "#import dlib\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import sys\n",
    "import time \n",
    "#import progressbar\n",
    "#from tqdm.notebook import tqdm_notebook\n",
    "from sklearn import svm, datasets\n",
    "from sklearn.dummy import DummyClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.utils import shuffle\n",
    "from sklearn.model_selection import train_test_split, GridSearchCV, learning_curve, ShuffleSplit\n",
    "from sklearn.metrics import classification_report, accuracy_score, confusion_matrix, recall_score, precision_score, f1_score\n",
    "from tensorflow.python.keras.applications.vgg16 import preprocess_input\n",
    "from tensorflow.python.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.python.keras.callbacks import EarlyStopping, ModelCheckpoint\n",
    "import keras\n",
    "from keras.utils import np_utils\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, AveragePooling2D, Conv2D, Flatten"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>face_shape</th>\n",
       "      <th>file_name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>4</td>\n",
       "      <td>0.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>1.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>2.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>3.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2</td>\n",
       "      <td>4.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9995</th>\n",
       "      <td>2</td>\n",
       "      <td>9995.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9996</th>\n",
       "      <td>3</td>\n",
       "      <td>9996.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9997</th>\n",
       "      <td>2</td>\n",
       "      <td>9997.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9998</th>\n",
       "      <td>2</td>\n",
       "      <td>9998.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9999</th>\n",
       "      <td>2</td>\n",
       "      <td>9999.png</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10000 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     face_shape file_name\n",
       "0             4     0.png\n",
       "1             4     1.png\n",
       "2             3     2.png\n",
       "3             0     3.png\n",
       "4             2     4.png\n",
       "...         ...       ...\n",
       "9995          2  9995.png\n",
       "9996          3  9996.png\n",
       "9997          2  9997.png\n",
       "9998          2  9998.png\n",
       "9999          2  9999.png\n",
       "\n",
       "[10000 rows x 2 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# extracting dataframe of cartoon dataset\n",
    "face_shape = pd.read_csv('D:/ELEC0134 Applied Machine Learning Systems/assignment/dataset_AMLS_20-21/cartoon_set/labels.csv', sep = '\\t')\n",
    "face_shape = face_shape.drop(columns = [face_shape.columns[0]]).drop(columns = [face_shape.columns[1]])\n",
    "face_shape['face_shape'] = face_shape['face_shape'].apply(str)\n",
    "face_shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     face_shape file_name\n",
      "6083          3  6083.png\n",
      "2771          0  2771.png\n",
      "5507          2  5507.png\n",
      "7866          3  7866.png\n",
      "6593          2  6593.png\n",
      "...         ...       ...\n",
      "1600          0  1600.png\n",
      "3682          3  3682.png\n",
      "4194          1  4194.png\n",
      "411           3   411.png\n",
      "6396          3  6396.png\n",
      "\n",
      "[9000 rows x 2 columns]      face_shape file_name\n",
      "6813          3  6813.png\n",
      "3489          2  3489.png\n",
      "8480          3  8480.png\n",
      "6951          1  6951.png\n",
      "6476          1  6476.png\n",
      "...         ...       ...\n",
      "7726          1  7726.png\n",
      "9624          1  9624.png\n",
      "4397          3  4397.png\n",
      "5966          1  5966.png\n",
      "9320          1  9320.png\n",
      "\n",
      "[1000 rows x 2 columns]\n"
     ]
    }
   ],
   "source": [
    "img = ('D:/ELEC0134 Applied Machine Learning Systems/assignment/dataset_AMLS_20-21/cartoon_set/img')\n",
    "# splitting training and testing dataset\n",
    "training, testing = np.split(face_shape.sample(frac=1), [int(0.9*len(face_shape)),]) #splitting at n-array\n",
    "print(training, testing)\n",
    "#print_summary(training, testing)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Dataset Preparation: \n",
      "Found 7200 validated image filenames belonging to 5 classes.\n",
      "\n",
      "Validation Dataset Preparation: \n",
      "Found 1800 validated image filenames belonging to 5 classes.\n"
     ]
    }
   ],
   "source": [
    "# set up data generator and further split training dataset for validation set\n",
    "data_generator = ImageDataGenerator(\n",
    "    rescale = 1./255.,\n",
    "    validation_split = 0.2,\n",
    "    horizontal_flip=True,\n",
    "    vertical_flip=True   \n",
    ")\n",
    "\n",
    "# Get batches of training dataset from the dataframe\n",
    "print(\"Training Dataset Preparation: \")\n",
    "train_generator = data_generator.flow_from_dataframe(\n",
    "        dataframe = training, directory = img ,\n",
    "        x_col = \"file_name\", y_col = \"face_shape\",\n",
    "        class_mode = 'categorical', target_size = (32,32),\n",
    "        batch_size = 128, subset = 'training') \n",
    "   \n",
    "# Get batches of validation dataset from the dataframe\n",
    "print(\"\\nValidation Dataset Preparation: \")\n",
    "validation_generator = data_generator.flow_from_dataframe(\n",
    "        dataframe = training, directory = img ,\n",
    "        x_col = \"file_name\", y_col = \"face_shape\",\n",
    "        class_mode = 'categorical', target_size = (32,32),\n",
    "        batch_size = 128, subset = 'validation')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential(name = 'Lenet5')\n",
    "\n",
    "model.add(Conv2D(filters=64, kernel_size=(3, 3), activation='relu', input_shape=(32,32,3)))\n",
    "model.add(AveragePooling2D())\n",
    "\n",
    "model.add(Conv2D(filters=128, kernel_size=(3, 3), activation='relu'))\n",
    "model.add(AveragePooling2D())\n",
    "\n",
    "\n",
    "model.add(Flatten())\n",
    "\n",
    "model.add(Dense(units=120, activation='relu'))\n",
    "\n",
    "model.add(Dense(units=84, activation='relu'))\n",
    "\n",
    "model.add(Dense(units=5, activation = 'softmax'))\n",
    "\n",
    "model.compile(loss = keras.losses.categorical_crossentropy,\n",
    "             optimizer = 'Adam', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"Lenet5\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_10 (Conv2D)           (None, 30, 30, 64)        1792      \n",
      "_________________________________________________________________\n",
      "average_pooling2d_9 (Average (None, 15, 15, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_11 (Conv2D)           (None, 13, 13, 128)       73856     \n",
      "_________________________________________________________________\n",
      "average_pooling2d_10 (Averag (None, 6, 6, 128)         0         \n",
      "_________________________________________________________________\n",
      "flatten_5 (Flatten)          (None, 4608)              0         \n",
      "_________________________________________________________________\n",
      "dense_13 (Dense)             (None, 120)               553080    \n",
      "_________________________________________________________________\n",
      "dense_14 (Dense)             (None, 84)                10164     \n",
      "_________________________________________________________________\n",
      "dense_15 (Dense)             (None, 5)                 425       \n",
      "=================================================================\n",
      "Total params: 639,317\n",
      "Trainable params: 639,317\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/13\n",
      "56/56 [==============================] - 66s 1s/step - loss: 1.6105 - accuracy: 0.2042 - val_loss: 1.6101 - val_accuracy: 0.1992\n",
      "Epoch 2/13\n",
      "56/56 [==============================] - 75s 1s/step - loss: 1.6097 - accuracy: 0.2022 - val_loss: 1.6090 - val_accuracy: 0.2159\n",
      "Epoch 3/13\n",
      "56/56 [==============================] - 51s 908ms/step - loss: 1.5891 - accuracy: 0.2551 - val_loss: 1.5176 - val_accuracy: 0.3666\n",
      "Epoch 4/13\n",
      "56/56 [==============================] - 42s 753ms/step - loss: 1.2782 - accuracy: 0.4143 - val_loss: 1.0980 - val_accuracy: 0.4976\n",
      "Epoch 5/13\n",
      "56/56 [==============================] - 51s 907ms/step - loss: 1.0634 - accuracy: 0.5208 - val_loss: 1.0952 - val_accuracy: 0.5203\n",
      "Epoch 6/13\n",
      "56/56 [==============================] - 39s 690ms/step - loss: 0.9099 - accuracy: 0.6201 - val_loss: 0.9403 - val_accuracy: 0.5927\n",
      "Epoch 7/13\n",
      "56/56 [==============================] - 38s 671ms/step - loss: 0.7873 - accuracy: 0.6838 - val_loss: 0.8071 - val_accuracy: 0.7273\n",
      "Epoch 8/13\n",
      "56/56 [==============================] - 37s 653ms/step - loss: 0.6653 - accuracy: 0.7574 - val_loss: 0.6388 - val_accuracy: 0.7566\n",
      "Epoch 9/13\n",
      "56/56 [==============================] - 36s 640ms/step - loss: 0.5570 - accuracy: 0.7999 - val_loss: 0.5181 - val_accuracy: 0.8224\n",
      "Epoch 10/13\n",
      "56/56 [==============================] - 36s 643ms/step - loss: 0.4713 - accuracy: 0.8380 - val_loss: 0.4578 - val_accuracy: 0.8547\n",
      "Epoch 11/13\n",
      "56/56 [==============================] - 47s 838ms/step - loss: 0.4025 - accuracy: 0.8657 - val_loss: 0.3382 - val_accuracy: 0.8822\n",
      "Epoch 12/13\n",
      "56/56 [==============================] - 62s 1s/step - loss: 0.3568 - accuracy: 0.8869 - val_loss: 0.2630 - val_accuracy: 0.8995\n",
      "Epoch 13/13\n",
      "56/56 [==============================] - 62s 1s/step - loss: 0.3255 - accuracy: 0.8978 - val_loss: 0.3709 - val_accuracy: 0.9073\n"
     ]
    }
   ],
   "source": [
    "# batch_size = 32\n",
    "history = model.fit_generator(\n",
    "                                train_generator,\n",
    "                                epochs=13,\n",
    "                                steps_per_epoch=train_generator.samples // train_generator.batch_size, \n",
    "                                validation_data=validation_generator,\n",
    "                                validation_steps=validation_generator.samples // validation_generator.batch_size\n",
    "                                )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x265bd074eb8>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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(history.history['accuracy'], 'o-', label='Training acc')\n",
    "plt.plot(history.history['val_accuracy'], 'x-', label='Validation acc')\n",
    "plt.xlabel('iterations')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x265bd113240>"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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(history.history['loss'], 'o-', label='Training Loss')\n",
    "plt.plot(history.history['val_loss'], 'x-', label='Validation Loss')\n",
    "plt.xlabel('iterations')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<keras.layers.convolutional.Conv2D at 0x265bf8da1d0>,\n",
       " <keras.layers.pooling.AveragePooling2D at 0x265bf8daba8>,\n",
       " <keras.layers.convolutional.Conv2D at 0x265bf8d7390>,\n",
       " <keras.layers.pooling.AveragePooling2D at 0x265bf8d7ac8>,\n",
       " <keras.layers.core.Flatten at 0x265bf8d7748>,\n",
       " <keras.layers.core.Dense at 0x265bf8d7eb8>,\n",
       " <keras.layers.core.Dense at 0x265bf8d7cc0>,\n",
       " <keras.layers.core.Dense at 0x265bfc02978>]"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "filter sizes :  (3, 3, 3, 64)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "D:\\Anaconda\\envs\\lab1\\lib\\site-packages\\ipykernel_launcher.py:13: MatplotlibDeprecationWarning: Adding an axes using the same arguments as a previous axes currently reuses the earlier instance.  In a future version, a new instance will always be created and returned.  Meanwhile, this warning can be suppressed, and the future behavior ensured, by passing a unique label to each axes instance.\n",
      "  del sys.path[0]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVAAAADnCAYAAABIUA6gAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/d3fzzAAAACXBIWXMAAAsTAAALEwEAmpwYAAARdElEQVR4nO3de2zO9//G8d4bZdQiZmNUnOpcc5hzOsMYihLHrXFYbM7UKU7FmFKLbcYcxjS2OXWWrKy6aqw11h3QiC3qWKSLremCBl10JtPf/7+019v98k5/h+/z8afL676v3Wtf+UTf/XwCpaWlIQCA4D3xP10AAP6vYoECgBELFACMWKAAYMQCBQCjSiqMjo6WP6LfsmWLfPEmTZrIvLS0NCD/QhACgYDs6jptkJWVJfNXXnnFW9f09HRZZuLEiXI+MzNT5pGRkV66zps3T/bMzc2V82FhYTJPSUnx9pn27dtXdnV9ZsnJyTJ//fXXvXXNzMyUXUtKSuR8y5YtZd6sWTNvXd955x3ZtaCgQM5v377d9RZeurq+/1euXCnnr1+/LvOkpKQye3IFCgBGLFAAMGKBAoARCxQAjFigAGDEAgUAo4A63lNYWCiPBsTGxsoXT0tLk3m1atW8Hbdo06aN7LpixQo5n5SUJPMjR45463rixAnZNSUlRc7n5eXJ/MCBA766yp7Tpk2Tw6NHj5Z57969K+wY29atW+X8oUOHZJ6enl5hx9hCQ0PlfJ06dWTetm1bb12/+uor2XXkyJFyPjs7W+ZRUVEV8rW6evVqOdy+fXuZDx48mGNMAOATCxQAjFigAGDEAgUAIxYoABixQAHAiAUKAEbydnYTJkyQw0ePHpV5amqqzGNiYmQeDNet04qKimQeGRnprYtLt27dZO46k3rw4EGPbcoXCOgjeoMGDZJ57dq1fdaR5s+fL/OcnByZd+zY0WcdKTo6Wuauz/3evXs+60iu29W5bgNZUV07dOgg8/Hjx8t848aNMh88eHCZf84VKAAYsUABwIgFCgBGLFAAMGKBAoARCxQAjFigAGAk7wcKACgfV6AAYMQCBQAjFigAGLFAAcBI3kwkxPGgpv3798th18PR9u/f7+3hVzNnzpRdi4uL5Xznzp1dr19hD0ALDw+X81evXpV5aGiol66LFy+WPZctW/ZYrx8WFlZhn+nUqVPl/OHDh2Wen5/vrWv//v1l14cPH8r5oUOHytzn12qIYwfcunVLDo8bN07mvh7Wt2nTJtkzLi5Ozl+8eFHmLVq04KFyAOATCxQAjFigAGDEAgUAIxYoABixQAHASB5jGjt2rBzu06ePzJ9//vngGxm1a9dO5pMmTZL52bNnfdaRrl+/LnPXMaatW7fKfPr06UF3KkuzZs1kXqNGDZlXrVpV5iUlJUF3Ks/bb78t8xkzZsg8NjbWWxeXI0eOyNx1f4rc3FyfdaSuXbvKPD09Xea9evXy2KZ8s2bNkvlvv/0m89DQUNP7cgUKAEYsUAAwYoECgBELFACMWKAAYMQCBQAjFigAGMlzoEVFRXL4l19+kfmmTZtkvmHDBpkH4+TJkzJPTEyUeX5+vsx9PjtqwIABMk9ISJB5kyZNvHVRXOcNCwsLZV6nTh2fdaSZM2fKfN++fTKfPXu2zzrS+++/L/NatWrJ/MaNGz7rSFlZWTLftWuXzBctWiTzhQsXBt2pLDt27JB5QUGBzF27rHHjxmX+OVegAGDEAgUAIxYoABixQAHAiAUKAEYsUAAwYoECgFHA5/lGAPhPwhUoABixQAHAiAUKAEYsUAAwkjcTCQQC8idMrptJfPnllzKfNWtWQP6F4Miu0dHRcvjhw4cyz8jI8NZ14sSJsmtYWJicHzJkiMz79evnpWuLFi1kz4sXL8r55s2byzwvL8/bZxodHS27VqlSRc4PHDhQ5pMnT/bWde3atbLr4MGD5XxeXp7Mhw8fXmHfV5cuXZLDrq+RoUOHeum6bNky2XPNmjVyPicnR+adOnUqsydXoABgxAIFACMWKAAYsUABwIgFCgBGLFAAMJK/Cx8fHy+PBrz55pvyxZs2bep8f9dfeFQHDhyQXRs0aCDnO3fuLPPS0lJvXZOTk2XX2NhYVxfXW3jpWq9ePflGrmczHTt2TOaVKlXy9plOnTpVdt2+fbucP378uMx79uxZYUeDdu7cKYdd33c+v1ZLHV9srmNMVatWlXmjRo28dC0uLpY9R40aJecHDRok8/KOXHIFCgBGLFAAMGKBAoARCxQAjFigAGDEAgUAIxYoABjJ29mNHDlSDicnJ8v8/v37Mk9ISJB5MHbt2iXzU6dOydx1Czmf4uPjZX748GGZZ2Zmyrxv375BdyrLiBEjZL5582aZV6okv7y82rZtm8ynTJki87p16/qsI50+fVrmrv+/PXv29FlHeuIJfY119OhRmdevX99nnXJt2rRJ5hkZGTJfuHCh6X25AgUAIxYoABixQAHAiAUKAEYsUAAwYoECgBELFACM5P1AAQDl4woUAIxYoABgxAIFACMWKAAYybs99OnTR/6EKSkpSb6462YT69evr7AHdW3cuFEOz5kzR7+4xwd1dejQQXZ13Szizp07Mv/ss8+8dJ02bZrs+e6777rmZb5v3z5vn2kgEJBdz549K+cLCgpk/uqrr3rreubMGdn1iy++kPOuh8o1b97cW9fff/9ddr19+7acd90YZ8GCBV66Hjp0SPasXbu2nHfdTCQ7O5uHygGATyxQADBigQKAEQsUAIxYoABgxAIFACP5u/Bz5syRRwPGjBkjX7x79+7O93f9hUcVHR0tu65Zs0bOv/feezL3eeTmzp07smtxcbGcDw8Pd72Fl65FRUWyZ1pampxv3LixzF966SVvn+nNmzdl11u3bsn5WbNmyfzIkSMVduSuVatWcvjXX3+VeWhoaIV1HTdunBz+5JNPZP7UU0956Xr+/HnZs3Xr1nI+Pz9f5o0aNeIYEwD4xAIFACMWKAAYsUABwIgFCgBGLFAAMGKBAoCRvJ1dRESEHL5586bMAwF9xMvn85h69+4t8xUrVsh8586d3rq4LF++XObJyckyd33uvj7XWrVqyXzChAkyP336tJcej8J1u7K4uDiZV69e3Wcd6eeff5Z5amqqzKtUqSJzn99Xru/hypUrP1bu6/uuTZs2Mg8LC5N5x44dZX78+PEy/5wrUAAwYoECgBELFACMWKAAYMQCBQAjFigAGLFAAcBI3g8UAFA+rkABwIgFCgBGLFAAMJK/Cx/ieB7K+vXr5fAzzzwj8wkTJlTYs1tWrlwph6tVqybzhQsXeusaCARkV9e/S7ueMzN58mQvXV0909PT5XxJSYnMhw8f7u0zTUpKkl1dz7y6fPmyzEtLS711TUxMlF337t0r513PGktKSvLWddu2bbLr1KlT5Xznzp1lnpOT46VrtWrVZM979+7J+T59+sj86NGjPBMJAHxigQKAEQsUAIxYoABgxAIFACMWKAAYsUABwEieA3U9D2XatGkyP3HihMxdz9QJRlpamszr1q0r827dunnr4tKlSxeZf/rppzLv0aOHzzrlys/Pl/mGDRtkvmrVKn9lHLZs2SLzS5cuyfzatWs+60iuz/X8+fMynzt3rsc2mut73HUWeOTIkT7rlGvAgAEyz8zMlLnrv7M8XIECgBELFACMWKAAYMQCBQAjFigAGLFAAcCIBQoARvIcaGpqqhxesGCBzF1ns3waMmSIzBMSEmR+5coVmbdv3z7YSuXKzs6W+aJFi2TuOkcYGRkZbKUyNWzYUOZNmjSRec2aNWX+77//BlupXCdPnpS565xwjRo1ZO76bw3G/fv3Zd6pUyeZv/XWW966uIwYMULmubm5Mh82bJjHNuWbPXu2zHv16iXzb775xvS+XIECgBELFACMWKAAYMQCBQAjFigAGLFAAcCIBQoARgHXM8gBAGXjChQAjFigAGDEAgUAI/m78P3795f/QDpjxgz54hkZGTLfunWrfuhSEPbs2SO7jh07Vs67flf22LFj3rp26NBBdr1w4YKc//vvv11v4aVrVFSU7On69/Pw8HCZ79+/39tnOm7cOFnmgw8+kPMREREyv3v3rreuu3btkl1dzwpbvHixzNeuXeut6+rVq2XXdevWyfmCggKZh4WFeem6Y8cO2fPu3bty3nV/gvj4+DJ7cgUKAEYsUAAwYoECgBELFACMWKAAYMQCBQAj1zEmObxkyRKZnzt3LvhGRkVFRTJPSUmR+ccff+yzjrR06VKZx8bGPla+b9++oDuV5YcffpD59OnTZb57924vPR7FqFGjZF5YWCjzuLg4n3Wk8ePHyzwrK0vmgYC3U0pOy5cvl/l3330n87CwMJ91yuU6+rd+/XqZr1ixwvS+XIECgBELFACMWKAAYMQCBQAjFigAGLFAAcCIBQoARvIcqOu82vz582W+Y8cOmU+aNEnmwbh27ZrMZ8+eLfP09HSZt2rVKuhO5fnpp59k/uDBA5k/99xz3rooK1eulHlCQoLMK/JsbU5OjsyHDh0q84p8tE1JSYnMhw8fLvOGDRv6rCPduHFD5t9++63MFy5cKHPX7fAe1YcffijzmjVryvyPP/4wvS9XoABgxAIFACMWKAAYsUABwIgFCgBGLFAAMGKBAoBRoCLPvwHA/ydcgQKAEQsUAIxYoABgxAIFACN5M5GQkBD5E6bo6Gg57LoZRZcuXbw9HSsQCMiurh+WXbx4UeYtW7b01rVfv36yTGZmppx33Yyke/fuXrp+/vnnsucbb7wh57Ozs2UeFRXl7TO9evWq7Hry5MnHev3Y2Nj/NV+rqampMo+JifHW9c8//5Rl7t+/L+cvX74s8759+3rp+rifabNmzWSel5dXZk+uQAHAiAUKAEYsUAAwYoECgBELFACMWKAAYCSPMRUUFMjhHj16yLxWrVrBNzJyHf35+uuvZd6+fXuPbTTXc2RefPFFmS9ZskTmx44dC7ZSmRYvXixz19GQzZs3yzwqKiroTuVZs2aNzD/66COZh4WFeeviMnfuXJkPHDhQ5vn5+TKPiYkJtlK5Lly4IPMaNWrI/M6dO966KF27dpX52bNnZb506VLT+3IFCgBGLFAAMGKBAoARCxQAjFigAGDEAgUAIxYoABjJc6D169eXw3v37pX5X3/9FXwjo6efflrmp0+flnm9evVk3rBhw6A7ladu3boyd90GsFu3bt66KFu2bJF527ZtZZ6VleWzjjR+/HiZu84rus7Ovvzyy8FWKldJSYnM4+LiZF6Rn2uvXr1kHgjou9F9//33HtuUz3XOt7i4WOYNGjQwvS9XoABgxAIFACMWKAAYsUABwIgFCgBGLFAAMGKBAoBRwHVPRwBA2bgCBQAjFigAGLFAAcCIBQoARvJmIocOHZI/YVqwYIF88QkTJsh8yZIl+k4EQQgEArKr60Fe4eHhMp83b563rnl5ebLr6NGj5fyZM2dcb+Gla0xMjOwZGRkp51944QWZv/baa94+0+PHj8uuP/74o5x3PVSstLTUW9fdu3fLrq4bdJw/f17miYmJ3rqGhITIro0bN5bDBw8elHm7du18dZU9r1y58lgvHhERUWZPrkABwIgFCgBGLFAAMGKBAoARCxQAjFigAGAkjzH1799fDhcUFMh80qRJwTcycv1Of+/evWVeUc8ZCgkJCUlKSpK561lD169fl7n1+S7/XYsWLWSenp4u88TERC89HsWpU6dkHh8fL3PX17pPrud3nTt3TuYV+bmuW7dO5tWrV5d5YWGhzNu1axd0p7Ls3r1b5q7/Dtczz9LS0sr8c65AAcCIBQoARixQADBigQKAEQsUAIxYoABgxAIFACN5DnTMmDFy+MCBAzJ33ZbL5/OYXGcrO3XqJHPXmUbXLeaCMWXKFJk3bdpU5g8fPpT5nj17gu5UlkGDBsk8IiJC5hX5/3/nzp0yf/LJJ2W+atUqmd++fTvYSuUaNmyYzFNSUmT+4MEDmVeuXDnYSuVatGiRzDMyMmT+zz//eOuiPPvsszLPzc2VeevWrU3vyxUoABixQAHAiAUKAEYsUAAwYoECgBELFACMWKAAYBTweRYPAP6TcAUKAEYsUAAwYoECgBELFACMWKAAYMQCBQCj/wI2BG5Hg4NGZAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 64 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# retrieve weights from the first convolutional layer\n",
    "filters, biases = model.layers[0].get_weights()\n",
    "print('filter sizes : ', filters.shape)\n",
    "\n",
    "# normalize filter values\n",
    "f_min, f_max = filters.min(), filters.max()\n",
    "filters = (filters - f_min) / (f_max - f_min)\n",
    "\n",
    "n_filters = filters.shape[3]\n",
    "for i in range(n_filters):\n",
    "    f = filters[:, :, :, i]\n",
    "    for j in range(f.shape[2]):\n",
    "        plt.subplot(8, 8, i+1)\n",
    "        plt.imshow(f[:, :, j], cmap='gray')\n",
    "        plt.axis('off')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# saving model\n",
    "from keras.models import load_model\n",
    "\n",
    "modelFile = 'face_shape.h5'\n",
    "\n",
    "my_model.save(modelFile)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'C:\\\\Users\\\\82419\\\\Machine Learning\\\\assignment\\\\task B'"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "os.getcwd()"
   ]
  }
 ],
 "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": "3.6.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}