02. Training ML Model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import the necessary packages\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import os\n",
    "\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.keras.applications import MobileNetV2\n",
    "from tensorflow.keras.layers import AveragePooling2D\n",
    "from tensorflow.keras.layers import Dropout\n",
    "from tensorflow.keras.layers import Flatten\n",
    "from tensorflow.keras.layers import Dense\n",
    "from tensorflow.keras.layers import Input\n",
    "from tensorflow.keras.models import Model\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "from sklearn.preprocessing import LabelBinarizer\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import classification_report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load preprocessed data stored in NumPy arrays\n",
    "data = np.load('data.npy')\n",
    "labels = np.load('labels.npy')\n",
    "\n",
    "# Declare hyperparameter constants\n",
    "    # Initial learning rate\n",
    "    # Epoch number\n",
    "    # Batch size\n",
    "INIT_LR = 1e-4\n",
    "EPOCHS = 20\n",
    "BS = 32"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Apply one-hot encoding on the labels\n",
    "lb = LabelBinarizer()\n",
    "labels = lb.fit_transform(labels)\n",
    "labels = to_categorical(labels)\n",
    "\n",
    "# Split data to two partitions: training set (80%) and testing set (20%)\n",
    "(trainX, testX, trainY, testY) = train_test_split(data, labels,\n",
    "\ttest_size=0.20, stratify=labels, random_state=42)\n",
    "\n",
    "# Prepare image generator for training data augmentation\n",
    "aug = ImageDataGenerator(\n",
    "    rotation_range=20,\n",
    "    zoom_range=0.15,\n",
    "    width_shift_range=0.2,\n",
    "    height_shift_range=0.2,\n",
    "    shear_range=0.15,\n",
    "    horizontal_flip=True,\n",
    "    fill_mode=\"nearest\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\lunesa\\anaconda3\\envs\\env-a\\lib\\site-packages\\keras_applications\\mobilenet_v2.py:294: UserWarning: `input_shape` is undefined or non-square, or `rows` is not in [96, 128, 160, 192, 224]. Weights for input shape (224, 224) will be loaded as the default.\n",
      "  warnings.warn('`input_shape` is undefined or non-square, '\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:sample_weight modes were coerced from\n",
      "  ...\n",
      "    to  \n",
      "  ['...']\n",
      "Train for 34 steps, validate on 276 samples\n",
      "Epoch 1/20\n",
      "34/34 [==============================] - 248s 7s/step - loss: 0.7129 - accuracy: 0.6320 - val_loss: 0.3795 - val_accuracy: 0.8594\n",
      "Epoch 2/20\n",
      "34/34 [==============================] - 248s 7s/step - loss: 0.4507 - accuracy: 0.7846 - val_loss: 0.2225 - val_accuracy: 0.9414\n",
      "Epoch 3/20\n",
      "34/34 [==============================] - 249s 7s/step - loss: 0.3243 - accuracy: 0.8642 - val_loss: 0.1616 - val_accuracy: 0.9609\n",
      "Epoch 4/20\n",
      "34/34 [==============================] - 287s 8s/step - loss: 0.2686 - accuracy: 0.8933 - val_loss: 0.1312 - val_accuracy: 0.9648\n",
      "Epoch 5/20\n",
      "34/34 [==============================] - 229s 7s/step - loss: 0.2196 - accuracy: 0.9157 - val_loss: 0.1025 - val_accuracy: 0.9648\n",
      "Epoch 6/20\n",
      "34/34 [==============================] - 218s 6s/step - loss: 0.2241 - accuracy: 0.9139 - val_loss: 0.0865 - val_accuracy: 0.9688\n",
      "Epoch 7/20\n",
      "34/34 [==============================] - 222s 7s/step - loss: 0.1798 - accuracy: 0.9316 - val_loss: 0.0738 - val_accuracy: 0.9727\n",
      "Epoch 8/20\n",
      "34/34 [==============================] - 226s 7s/step - loss: 0.1751 - accuracy: 0.9354 - val_loss: 0.0717 - val_accuracy: 0.9844\n",
      "Epoch 9/20\n",
      "34/34 [==============================] - 224s 7s/step - loss: 0.1499 - accuracy: 0.9476 - val_loss: 0.0587 - val_accuracy: 0.9805\n",
      "Epoch 10/20\n",
      "34/34 [==============================] - 221s 6s/step - loss: 0.1509 - accuracy: 0.9382 - val_loss: 0.0575 - val_accuracy: 0.9805\n",
      "Epoch 11/20\n",
      "34/34 [==============================] - 220s 6s/step - loss: 0.1335 - accuracy: 0.9466 - val_loss: 0.0548 - val_accuracy: 0.9805\n",
      "Epoch 12/20\n",
      "34/34 [==============================] - 221s 7s/step - loss: 0.1279 - accuracy: 0.9457 - val_loss: 0.0613 - val_accuracy: 0.9844\n",
      "Epoch 13/20\n",
      "34/34 [==============================] - 224s 7s/step - loss: 0.1122 - accuracy: 0.9551 - val_loss: 0.0480 - val_accuracy: 0.9844\n",
      "Epoch 14/20\n",
      "34/34 [==============================] - 222s 7s/step - loss: 0.1033 - accuracy: 0.9691 - val_loss: 0.0669 - val_accuracy: 0.9844\n",
      "Epoch 15/20\n",
      "34/34 [==============================] - 221s 7s/step - loss: 0.1061 - accuracy: 0.9597 - val_loss: 0.0464 - val_accuracy: 0.9844\n",
      "Epoch 16/20\n",
      "34/34 [==============================] - 229s 7s/step - loss: 0.0954 - accuracy: 0.9644 - val_loss: 0.0465 - val_accuracy: 0.9844\n",
      "Epoch 17/20\n",
      "34/34 [==============================] - 229s 7s/step - loss: 0.1022 - accuracy: 0.9635 - val_loss: 0.0588 - val_accuracy: 0.9844\n",
      "Epoch 18/20\n",
      "34/34 [==============================] - 229s 7s/step - loss: 0.1015 - accuracy: 0.9625 - val_loss: 0.0516 - val_accuracy: 0.9844\n",
      "Epoch 19/20\n",
      "34/34 [==============================] - 214s 6s/step - loss: 0.0847 - accuracy: 0.9682 - val_loss: 0.0496 - val_accuracy: 0.9844\n",
      "Epoch 20/20\n",
      "34/34 [==============================] - 218s 6s/step - loss: 0.0917 - accuracy: 0.9616 - val_loss: 0.0509 - val_accuracy: 0.9844\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "   with_mask       0.97      1.00      0.99       138\n",
      "without_mask       1.00      0.97      0.99       138\n",
      "\n",
      "    accuracy                           0.99       276\n",
      "   macro avg       0.99      0.99      0.99       276\n",
      "weighted avg       0.99      0.99      0.99       276\n",
      "\n"
     ]
    }
   ],
   "source": [
    "imageSize=224\n",
    "\n",
    "# Load the MobileNetV2 network with pre-trained weights (imagenet)\n",
    "# and excluding head network layer\n",
    "baseModel = MobileNetV2(weights=\"imagenet\", include_top=False,\n",
    "    input_tensor=Input(shape=(imageSize, imageSize, 3)))\n",
    "\n",
    "# Build a new head network layer\n",
    "headModel = baseModel.output\n",
    "headModel = AveragePooling2D(pool_size=(7, 7))(headModel)\n",
    "headModel = Flatten(name=\"flatten\")(headModel)\n",
    "headModel = Dense(128, activation=\"relu\")(headModel)\n",
    "headModel = Dropout(0.5)(headModel)\n",
    "headModel = Dense(2, activation=\"softmax\")(headModel)\n",
    "\n",
    "# Prepend new head Fully Connected model to the base model\n",
    "model = Model(inputs=baseModel.input, outputs=headModel)\n",
    "\n",
    "# Loop over layers of base model and make sure they will not be\n",
    "# trained or updated during the backpropagation process\n",
    "for layer in baseModel.layers:\n",
    "    layer.trainable = False\n",
    "    \n",
    "# Compile the model\n",
    "opt = Adam(lr=INIT_LR, decay=INIT_LR / EPOCHS)\n",
    "model.compile(loss=\"binary_crossentropy\", optimizer=opt,\n",
    "    metrics=[\"accuracy\"])\n",
    "\n",
    "# Train the head layer of the model\n",
    "H = model.fit(\n",
    "    aug.flow(trainX, trainY, batch_size=BS),\n",
    "    steps_per_epoch=len(trainX) // BS,\n",
    "    validation_data=(testX, testY),\n",
    "    validation_steps=len(testX) // BS,\n",
    "    epochs=EPOCHS)\n",
    "\n",
    "# Make predictions on the testing set\n",
    "predIdxs = model.predict(testX, batch_size=BS)\n",
    "\n",
    "# For each image in the testing set, find the index of \n",
    "# the label with corresponding highest predicted probability\n",
    "predIdxs = np.argmax(predIdxs, axis=1)\n",
    "\n",
    "# Display classification report\n",
    "print(classification_report(testY.argmax(axis=1), predIdxs,\n",
    "    target_names=lb.classes_))\n",
    "\n",
    "# Serialize the model to file\n",
    "model.save('mask_detector.model', save_format=\"h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot training and testing accuracy\n",
    "N = EPOCHS\n",
    "plt.style.use(\"ggplot\")\n",
    "plt.figure()\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_acc\")\n",
    "plt.plot(np.arange(0, N), H.history[\"val_accuracy\"], label=\"val_acc\")\n",
    "plt.title(\"Training Loss and Accuracy\")\n",
    "plt.xlabel(\"Epoch #\")\n",
    "plt.ylabel(\"Loss/Accuracy\")\n",
    "plt.legend(loc=\"lower left\")\n",
    "plt.savefig('plot.png')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}