train_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From <ipython-input-42-c594da6aa825>:121: Model.fit_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use Model.fit, which supports generators.\n",
      "Epoch 1/100\n",
      "28/28 [==============================] - 62s 2s/step - loss: 0.7769 - accuracy: 0.7271 - val_loss: 0.7551 - val_accuracy: 0.4978\n",
      "Epoch 2/100\n",
      "28/28 [==============================] - 62s 2s/step - loss: 0.4632 - accuracy: 0.8175 - val_loss: 0.6621 - val_accuracy: 0.6688\n",
      "Epoch 3/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.4164 - accuracy: 0.8366 - val_loss: 0.8830 - val_accuracy: 0.5022\n",
      "Epoch 4/100\n",
      "28/28 [==============================] - 66s 2s/step - loss: 0.3336 - accuracy: 0.8776 - val_loss: 0.6399 - val_accuracy: 0.6797\n",
      "Epoch 5/100\n",
      "28/28 [==============================] - 69s 2s/step - loss: 0.3076 - accuracy: 0.8905 - val_loss: 0.7075 - val_accuracy: 0.5043\n",
      "Epoch 6/100\n",
      "28/28 [==============================] - 70s 2s/step - loss: 0.2765 - accuracy: 0.8984 - val_loss: 0.8473 - val_accuracy: 0.5108\n",
      "Epoch 7/100\n",
      "28/28 [==============================] - 65s 2s/step - loss: 0.3015 - accuracy: 0.8827 - val_loss: 0.7622 - val_accuracy: 0.5281\n",
      "Epoch 8/100\n",
      "28/28 [==============================] - 63s 2s/step - loss: 0.2611 - accuracy: 0.8989 - val_loss: 0.5041 - val_accuracy: 0.8398\n",
      "Epoch 9/100\n",
      "28/28 [==============================] - 67s 2s/step - loss: 0.2255 - accuracy: 0.9118 - val_loss: 2.3198 - val_accuracy: 0.5022\n",
      "Epoch 10/100\n",
      "28/28 [==============================] - 63s 2s/step - loss: 0.2326 - accuracy: 0.9152 - val_loss: 0.7585 - val_accuracy: 0.5909\n",
      "Epoch 11/100\n",
      "28/28 [==============================] - 67s 2s/step - loss: 0.2064 - accuracy: 0.9220 - val_loss: 2.4014 - val_accuracy: 0.5022\n",
      "Epoch 12/100\n",
      "28/28 [==============================] - 65s 2s/step - loss: 0.1933 - accuracy: 0.9304 - val_loss: 0.5404 - val_accuracy: 0.7814\n",
      "Epoch 13/100\n",
      "28/28 [==============================] - 62s 2s/step - loss: 0.2024 - accuracy: 0.9287 - val_loss: 0.8687 - val_accuracy: 0.6277\n",
      "Epoch 14/100\n",
      "28/28 [==============================] - 62s 2s/step - loss: 0.1835 - accuracy: 0.9248 - val_loss: 0.9152 - val_accuracy: 0.6710\n",
      "Epoch 15/100\n",
      "28/28 [==============================] - 62s 2s/step - loss: 0.1584 - accuracy: 0.9394 - val_loss: 0.3055 - val_accuracy: 0.8701\n",
      "Epoch 16/100\n",
      "28/28 [==============================] - 64s 2s/step - loss: 0.1460 - accuracy: 0.9394 - val_loss: 0.3580 - val_accuracy: 0.8485\n",
      "Epoch 17/100\n",
      "28/28 [==============================] - 88s 3s/step - loss: 0.1535 - accuracy: 0.9416 - val_loss: 0.5151 - val_accuracy: 0.8139\n",
      "Epoch 18/100\n",
      "28/28 [==============================] - 82s 3s/step - loss: 0.1536 - accuracy: 0.9439 - val_loss: 0.1235 - val_accuracy: 0.9459\n",
      "Epoch 19/100\n",
      "28/28 [==============================] - 88s 3s/step - loss: 0.1279 - accuracy: 0.9433 - val_loss: 0.4818 - val_accuracy: 0.8247\n",
      "Epoch 20/100\n",
      "28/28 [==============================] - 77s 3s/step - loss: 0.1196 - accuracy: 0.9528 - val_loss: 0.2126 - val_accuracy: 0.9199\n",
      "Epoch 21/100\n",
      "28/28 [==============================] - 71s 3s/step - loss: 0.1207 - accuracy: 0.9545 - val_loss: 0.8774 - val_accuracy: 0.7424\n",
      "Epoch 22/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.1230 - accuracy: 0.9534 - val_loss: 0.2465 - val_accuracy: 0.9329\n",
      "Epoch 23/100\n",
      "28/28 [==============================] - 59s 2s/step - loss: 0.1362 - accuracy: 0.9483 - val_loss: 0.1451 - val_accuracy: 0.9567\n",
      "Epoch 24/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0918 - accuracy: 0.9652 - val_loss: 0.2613 - val_accuracy: 0.9242\n",
      "Epoch 25/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0879 - accuracy: 0.9635 - val_loss: 0.9540 - val_accuracy: 0.7489\n",
      "Epoch 26/100\n",
      "28/28 [==============================] - 73s 3s/step - loss: 0.1007 - accuracy: 0.9629 - val_loss: 0.1386 - val_accuracy: 0.9610\n",
      "Epoch 27/100\n",
      "28/28 [==============================] - 67s 2s/step - loss: 0.1030 - accuracy: 0.9607 - val_loss: 0.8150 - val_accuracy: 0.7879\n",
      "Epoch 28/100\n",
      "28/28 [==============================] - 59s 2s/step - loss: 0.0770 - accuracy: 0.9708 - val_loss: 0.1292 - val_accuracy: 0.9502\n",
      "Epoch 29/100\n",
      "28/28 [==============================] - 62s 2s/step - loss: 0.0968 - accuracy: 0.9691 - val_loss: 0.1072 - val_accuracy: 0.9545\n",
      "Epoch 30/100\n",
      "28/28 [==============================] - 62s 2s/step - loss: 0.1225 - accuracy: 0.9579 - val_loss: 0.1455 - val_accuracy: 0.9394\n",
      "Epoch 31/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0848 - accuracy: 0.9663 - val_loss: 0.4566 - val_accuracy: 0.8680\n",
      "Epoch 32/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0980 - accuracy: 0.9579 - val_loss: 0.1134 - val_accuracy: 0.9632\n",
      "Epoch 33/100\n",
      "28/28 [==============================] - 62s 2s/step - loss: 0.0864 - accuracy: 0.9719 - val_loss: 0.1085 - val_accuracy: 0.9610\n",
      "Epoch 34/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0856 - accuracy: 0.9669 - val_loss: 0.1763 - val_accuracy: 0.9459\n",
      "Epoch 35/100\n",
      "28/28 [==============================] - 67s 2s/step - loss: 0.0791 - accuracy: 0.9663 - val_loss: 0.1802 - val_accuracy: 0.9394\n",
      "Epoch 36/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0863 - accuracy: 0.9674 - val_loss: 0.3028 - val_accuracy: 0.9026\n",
      "Epoch 37/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0679 - accuracy: 0.9714 - val_loss: 0.2171 - val_accuracy: 0.9329\n",
      "Epoch 38/100\n",
      "28/28 [==============================] - 75s 3s/step - loss: 0.0766 - accuracy: 0.9691 - val_loss: 0.1142 - val_accuracy: 0.9632\n",
      "Epoch 39/100\n",
      "28/28 [==============================] - 68s 2s/step - loss: 0.0746 - accuracy: 0.9747 - val_loss: 0.2203 - val_accuracy: 0.9351\n",
      "Epoch 40/100\n",
      "28/28 [==============================] - 64s 2s/step - loss: 0.0582 - accuracy: 0.9764 - val_loss: 0.5004 - val_accuracy: 0.8745\n",
      "Epoch 41/100\n",
      "28/28 [==============================] - 63s 2s/step - loss: 0.0708 - accuracy: 0.9753 - val_loss: 0.1768 - val_accuracy: 0.9481\n",
      "Epoch 42/100\n",
      "28/28 [==============================] - 65s 2s/step - loss: 0.0639 - accuracy: 0.9736 - val_loss: 0.1635 - val_accuracy: 0.9481\n",
      "Epoch 43/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0660 - accuracy: 0.9753 - val_loss: 1.6409 - val_accuracy: 0.7035\n",
      "Epoch 44/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0996 - accuracy: 0.9618 - val_loss: 0.3158 - val_accuracy: 0.8831\n",
      "Epoch 45/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0991 - accuracy: 0.9590 - val_loss: 0.5036 - val_accuracy: 0.8506\n",
      "Epoch 46/100\n",
      "28/28 [==============================] - 62s 2s/step - loss: 0.0758 - accuracy: 0.9736 - val_loss: 1.2065 - val_accuracy: 0.7121\n",
      "Epoch 47/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.1061 - accuracy: 0.9596 - val_loss: 0.3318 - val_accuracy: 0.8680\n",
      "Epoch 48/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0725 - accuracy: 0.9725 - val_loss: 0.1839 - val_accuracy: 0.9286\n",
      "Epoch 49/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0677 - accuracy: 0.9747 - val_loss: 0.0719 - val_accuracy: 0.9632\n",
      "Epoch 50/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0617 - accuracy: 0.9730 - val_loss: 0.1926 - val_accuracy: 0.9264\n",
      "Epoch 51/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0639 - accuracy: 0.9742 - val_loss: 0.0594 - val_accuracy: 0.9848\n",
      "Epoch 52/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0525 - accuracy: 0.9820 - val_loss: 0.1964 - val_accuracy: 0.9394\n",
      "Epoch 53/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0561 - accuracy: 0.9837 - val_loss: 0.0850 - val_accuracy: 0.9697\n",
      "Epoch 54/100\n",
      "28/28 [==============================] - 67s 2s/step - loss: 0.0503 - accuracy: 0.9799 - val_loss: 0.0915 - val_accuracy: 0.9697\n",
      "Epoch 55/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0559 - accuracy: 0.9775 - val_loss: 0.1223 - val_accuracy: 0.9610\n",
      "Epoch 56/100\n",
      "28/28 [==============================] - 67s 2s/step - loss: 0.0614 - accuracy: 0.9764 - val_loss: 0.4499 - val_accuracy: 0.8810\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 57/100\n",
      "28/28 [==============================] - 86s 3s/step - loss: 0.0583 - accuracy: 0.9816 - val_loss: 0.0937 - val_accuracy: 0.9697\n",
      "Epoch 58/100\n",
      "28/28 [==============================] - 69s 2s/step - loss: 0.0686 - accuracy: 0.9753 - val_loss: 0.0672 - val_accuracy: 0.9762\n",
      "Epoch 59/100\n",
      "28/28 [==============================] - 84s 3s/step - loss: 0.0574 - accuracy: 0.9792 - val_loss: 0.0739 - val_accuracy: 0.9784\n",
      "Epoch 60/100\n",
      "28/28 [==============================] - 83s 3s/step - loss: 0.0701 - accuracy: 0.9708 - val_loss: 0.1155 - val_accuracy: 0.9589\n",
      "Epoch 61/100\n",
      "28/28 [==============================] - 81s 3s/step - loss: 0.0719 - accuracy: 0.9719 - val_loss: 0.1379 - val_accuracy: 0.9545\n",
      "Epoch 62/100\n",
      "28/28 [==============================] - 83s 3s/step - loss: 0.0508 - accuracy: 0.9820 - val_loss: 0.4099 - val_accuracy: 0.8831\n",
      "Epoch 63/100\n",
      "28/28 [==============================] - 67s 2s/step - loss: 0.0520 - accuracy: 0.9798 - val_loss: 0.1696 - val_accuracy: 0.9481\n",
      "Epoch 64/100\n",
      "28/28 [==============================] - 63s 2s/step - loss: 0.0592 - accuracy: 0.9775 - val_loss: 0.1344 - val_accuracy: 0.9632\n",
      "Epoch 65/100\n",
      "28/28 [==============================] - 63s 2s/step - loss: 0.0488 - accuracy: 0.9815 - val_loss: 0.1216 - val_accuracy: 0.9675\n",
      "Epoch 66/100\n",
      "28/28 [==============================] - 63s 2s/step - loss: 0.0405 - accuracy: 0.9855 - val_loss: 0.1439 - val_accuracy: 0.9524\n",
      "Epoch 67/100\n",
      "28/28 [==============================] - 67s 2s/step - loss: 0.0488 - accuracy: 0.9826 - val_loss: 1.4368 - val_accuracy: 0.7251\n",
      "Epoch 68/100\n",
      "28/28 [==============================] - 70s 3s/step - loss: 0.0735 - accuracy: 0.9742 - val_loss: 1.1942 - val_accuracy: 0.7532\n",
      "Epoch 69/100\n",
      "28/28 [==============================] - 63s 2s/step - loss: 0.0660 - accuracy: 0.9764 - val_loss: 0.4398 - val_accuracy: 0.8745\n",
      "Epoch 70/100\n",
      "28/28 [==============================] - 70s 3s/step - loss: 0.0487 - accuracy: 0.9815 - val_loss: 0.2187 - val_accuracy: 0.9351\n",
      "Epoch 71/100\n",
      "28/28 [==============================] - 71s 3s/step - loss: 0.0566 - accuracy: 0.9764 - val_loss: 0.0918 - val_accuracy: 0.9632\n",
      "Epoch 72/100\n",
      "28/28 [==============================] - 74s 3s/step - loss: 0.0595 - accuracy: 0.9787 - val_loss: 0.0709 - val_accuracy: 0.9784\n",
      "Epoch 73/100\n",
      "28/28 [==============================] - 62s 2s/step - loss: 0.0492 - accuracy: 0.9798 - val_loss: 0.1018 - val_accuracy: 0.9632\n",
      "Epoch 74/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0382 - accuracy: 0.9893 - val_loss: 0.4619 - val_accuracy: 0.8723\n",
      "Epoch 75/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0638 - accuracy: 0.9798 - val_loss: 0.0845 - val_accuracy: 0.9719\n",
      "Epoch 76/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0408 - accuracy: 0.9854 - val_loss: 0.2297 - val_accuracy: 0.9697\n",
      "Epoch 77/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0490 - accuracy: 0.9821 - val_loss: 0.1039 - val_accuracy: 0.9719\n",
      "Epoch 78/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0521 - accuracy: 0.9815 - val_loss: 0.0890 - val_accuracy: 0.9740\n",
      "Epoch 79/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0522 - accuracy: 0.9798 - val_loss: 0.0560 - val_accuracy: 0.9784\n",
      "Epoch 80/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0451 - accuracy: 0.9848 - val_loss: 0.1988 - val_accuracy: 0.9437\n",
      "Epoch 81/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0329 - accuracy: 0.9865 - val_loss: 0.3713 - val_accuracy: 0.9113\n",
      "Epoch 82/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0375 - accuracy: 0.9860 - val_loss: 0.0844 - val_accuracy: 0.9740\n",
      "Epoch 83/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0487 - accuracy: 0.9809 - val_loss: 0.0653 - val_accuracy: 0.9762\n",
      "Epoch 84/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0504 - accuracy: 0.9815 - val_loss: 0.2088 - val_accuracy: 0.9437\n",
      "Epoch 85/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0446 - accuracy: 0.9848 - val_loss: 0.1146 - val_accuracy: 0.9610\n",
      "Epoch 86/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0477 - accuracy: 0.9865 - val_loss: 0.5406 - val_accuracy: 0.8896\n",
      "Epoch 87/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0635 - accuracy: 0.9764 - val_loss: 0.1960 - val_accuracy: 0.9545\n",
      "Epoch 88/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0545 - accuracy: 0.9803 - val_loss: 0.4570 - val_accuracy: 0.8788\n",
      "Epoch 89/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0339 - accuracy: 0.9860 - val_loss: 0.3403 - val_accuracy: 0.9221\n",
      "Epoch 90/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0439 - accuracy: 0.9860 - val_loss: 0.0781 - val_accuracy: 0.9784\n",
      "Epoch 91/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0493 - accuracy: 0.9848 - val_loss: 0.5429 - val_accuracy: 0.8506\n",
      "Epoch 92/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0388 - accuracy: 0.9876 - val_loss: 0.2633 - val_accuracy: 0.9437\n",
      "Epoch 93/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0468 - accuracy: 0.9826 - val_loss: 0.1025 - val_accuracy: 0.9632\n",
      "Epoch 94/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0632 - accuracy: 0.9730 - val_loss: 0.0920 - val_accuracy: 0.9654\n",
      "Epoch 95/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0413 - accuracy: 0.9854 - val_loss: 0.4757 - val_accuracy: 0.8745\n",
      "Epoch 96/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0445 - accuracy: 0.9815 - val_loss: 0.1259 - val_accuracy: 0.9545\n",
      "Epoch 97/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0409 - accuracy: 0.9848 - val_loss: 0.1655 - val_accuracy: 0.9545\n",
      "Epoch 98/100\n",
      "28/28 [==============================] - 61s 2s/step - loss: 0.0317 - accuracy: 0.9860 - val_loss: 0.0806 - val_accuracy: 0.9654\n",
      "Epoch 99/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0461 - accuracy: 0.9826 - val_loss: 0.0965 - val_accuracy: 0.9654\n",
      "Epoch 100/100\n",
      "28/28 [==============================] - 60s 2s/step - loss: 0.0432 - accuracy: 0.9843 - val_loss: 0.0890 - val_accuracy: 0.9697\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tensorflow.keras.preprocessing.image import img_to_array\n",
    "from tensorflow.keras.utils import to_categorical, plot_model\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import BatchNormalization, Conv2D, MaxPooling2D, Activation, Flatten, Dropout, Dense\n",
    "from tensorflow.keras import backend as K\n",
    "from sklearn.model_selection import train_test_split\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import random\n",
    "import cv2\n",
    "import os\n",
    "import glob\n",
    "\n",
    "# initial parameters\n",
    "epochs = 100\n",
    "lr = 1e-3\n",
    "batch_size = 64\n",
    "img_dims = (96,96,3)\n",
    "\n",
    "data = []\n",
    "labels = []\n",
    "\n",
    "# load image files from the dataset\n",
    "image_files = [f for f in glob.glob(r'PATH_to_image_data' + \"/**/*\", recursive=True) if not os.path.isdir(f)]\n",
    "random.shuffle(image_files)\n",
    "\n",
    "# converting images to arrays and labelling the categories\n",
    "for img in image_files:\n",
    "\n",
    "    image = cv2.imread(img)\n",
    "    \n",
    "    image = cv2.resize(image, (img_dims[0],img_dims[1]))\n",
    "    image = img_to_array(image)\n",
    "    data.append(image)\n",
    "\n",
    "    label = img.split(os.path.sep)[-2] # C:\\Files\\gender_dataset_face\\woman\\face_1162.jpg\n",
    "    if label == \"woman\":\n",
    "        label = 1\n",
    "    else:\n",
    "        label = 0\n",
    "        \n",
    "    labels.append([label]) # [[1], [0], [0], ...]\n",
    "\n",
    "# pre-processing\n",
    "data = np.array(data, dtype=\"float\") / 255.0\n",
    "labels = np.array(labels)\n",
    "\n",
    "# split dataset for training and validation\n",
    "(trainX, testX, trainY, testY) = train_test_split(data, labels, test_size=0.2,\n",
    "                                                  random_state=42)\n",
    "\n",
    "trainY = to_categorical(trainY, num_classes=2) # [[1, 0], [0, 1], [0, 1], ...]\n",
    "testY = to_categorical(testY, num_classes=2)\n",
    "\n",
    "# augmenting datset \n",
    "aug = ImageDataGenerator(rotation_range=25, width_shift_range=0.1,\n",
    "                         height_shift_range=0.1, shear_range=0.2, zoom_range=0.2,\n",
    "                         horizontal_flip=True, fill_mode=\"nearest\")\n",
    "\n",
    "# define model\n",
    "def build(width, height, depth, classes):\n",
    "    model = Sequential()\n",
    "    inputShape = (height, width, depth)\n",
    "    chanDim = -1\n",
    "\n",
    "    if K.image_data_format() == \"channels_first\": #Returns a string, either 'channels_first' or 'channels_last'\n",
    "        inputShape = (depth, height, width)\n",
    "        chanDim = 1\n",
    "    \n",
    "    # The axis that should be normalized, after a Conv2D layer with data_format=\"channels_first\", \n",
    "    # set axis=1 in BatchNormalization.\n",
    "\n",
    "    model.add(Conv2D(32, (3,3), padding=\"same\", input_shape=inputShape))\n",
    "    model.add(Activation(\"relu\"))\n",
    "    model.add(BatchNormalization(axis=chanDim))\n",
    "    model.add(MaxPooling2D(pool_size=(3,3)))\n",
    "    model.add(Dropout(0.25))\n",
    "\n",
    "    model.add(Conv2D(64, (2,2), padding=\"same\"))\n",
    "    model.add(Activation(\"relu\"))\n",
    "    model.add(BatchNormalization(axis=chanDim))\n",
    "\n",
    "    model.add(Conv2D(64, (2,2), padding=\"same\"))\n",
    "    model.add(Activation(\"relu\"))\n",
    "    model.add(BatchNormalization(axis=chanDim))\n",
    "    model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "    model.add(Dropout(0.25))\n",
    "\n",
    "    model.add(Conv2D(128, (2,2), padding=\"same\"))\n",
    "    model.add(Activation(\"relu\"))\n",
    "    model.add(BatchNormalization(axis=chanDim))\n",
    "\n",
    "    model.add(Conv2D(128, (2,2), padding=\"same\"))\n",
    "    model.add(Activation(\"relu\"))\n",
    "    model.add(BatchNormalization(axis=chanDim))\n",
    "    model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "    model.add(Dropout(0.25))\n",
    "\n",
    "    model.add(Flatten())\n",
    "    model.add(Dense(1024))\n",
    "    model.add(Activation(\"relu\"))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(Dropout(0.5))\n",
    "\n",
    "    model.add(Dense(classes))\n",
    "    model.add(Activation(\"sigmoid\"))\n",
    "\n",
    "    return model\n",
    "\n",
    "# build model\n",
    "model = build(width=img_dims[0], height=img_dims[1], depth=img_dims[2],\n",
    "                            classes=2)\n",
    "\n",
    "# compile the model\n",
    "opt = Adam(lr=lr, decay=lr/epochs)\n",
    "model.compile(loss=\"binary_crossentropy\", optimizer=opt, metrics=[\"accuracy\"])\n",
    "\n",
    "# train the model\n",
    "H = model.fit_generator(aug.flow(trainX, trainY, batch_size=batch_size),\n",
    "                        validation_data=(testX,testY),\n",
    "                        steps_per_epoch=len(trainX) // batch_size,\n",
    "                        epochs=epochs, verbose=1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From c:\\users\\91982\\appdata\\local\\programs\\python\\python38\\lib\\site-packages\\tensorflow\\python\\training\\tracking\\tracking.py:111: Model.state_updates (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "This property should not be used in TensorFlow 2.0, as updates are applied automatically.\n",
      "WARNING:tensorflow:From c:\\users\\91982\\appdata\\local\\programs\\python\\python38\\lib\\site-packages\\tensorflow\\python\\training\\tracking\\tracking.py:111: Layer.updates (from tensorflow.python.keras.engine.base_layer) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "This property should not be used in TensorFlow 2.0, as updates are applied automatically.\n",
      "INFO:tensorflow:Assets written to: gender_detection.model\\assets\n"
     ]
    }
   ],
   "source": [
    "model.save('gender_detection.model')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1c2480f1430>"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.style.use(\"ggplot\")\n",
    "plt.figure()\n",
    "N = epochs\n",
    "plt.plot(np.arange(0,N), H.history[\"loss\"], label=\"train_loss\")\n",
    "plt.plot(np.arange(0,N), H.history[\"val_loss\"], label=\"val_loss\")\n",
    "plt.plot(np.arange(0,N), H.history[\"accuracy\"], label=\"train_accuracy\")\n",
    "plt.plot(np.arange(0,N), H.history[\"val_accuracy\"], label=\"val_accuracy\")\n",
    "\n",
    "plt.title(\"Training Loss and Accuracy\")\n",
    "plt.xlabel(\"Epoch #\")\n",
    "plt.ylabel(\"Loss/Accuracy\")\n",
    "plt.legend(loc=\"upper right\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.savefig('plot.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "Untitled17.ipynb",
   "provenance": []
  },
  "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}