train-model.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predict Covid-19 in X-ray images"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Part 2: Train the model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 1: import everything"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.keras.applications import VGG16\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\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from imutils import paths\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import cv2\n",
    "import os"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 2: define some variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define some paths\n",
    "dataset = './dataset'\n",
    "plot_path = './plot.png'\n",
    "model_save_path = 'covid19.model'\n",
    "\n",
    "# initialize the initial learning rate, number of epochs to train for, and batch size\n",
    "INIT_LR = 1e-3\n",
    "EPOCHS = 5\n",
    "BS = 8"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 3: load in the images from the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading images...\n",
      "Loaded 112 images!\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# grab the list of images in our dataset directory, then initialize\n",
    "# the list of data (i.e., images) and class images\n",
    "print(\"Loading images...\")\n",
    "image_paths = list(paths.list_images(dataset))\n",
    "data = []\n",
    "labels = []\n",
    "\n",
    "# loop over the image paths\n",
    "for image_path in image_paths:\n",
    "    # extract the class label from the filename\n",
    "    label = image_path.split(os.path.sep)[-2]\n",
    "\n",
    "    # load the image, swap color channels, and resize it to be a fixed\n",
    "    # 224x224 pixels while ignoring aspect ratio\n",
    "    image = cv2.imread(image_path)\n",
    "    image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n",
    "    image = cv2.resize(image, (224, 224))\n",
    "\n",
    "    # update the data and labels lists, respectively\n",
    "    data.append(image)\n",
    "    labels.append(label)\n",
    "\n",
    "print(\"Loaded {} images!\\n\".format(len(image_paths)))  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 4: do some preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert the data and labels to NumPy arrays while scaling the pixel\n",
    "# intensities to the range [0, 255]\n",
    "data = np.array(data) / 255.0\n",
    "labels = np.array(labels)\n",
    "\n",
    "# perform one-hot encoding on the labels\n",
    "lb = LabelBinarizer()\n",
    "labels = lb.fit_transform(labels)\n",
    "labels = to_categorical(labels)\n",
    "\n",
    "# partition the data into training and testing splits using 80% of\n",
    "# the data for training and the remaining 20% for testing\n",
    "(trainX, testX, trainY, testY) = train_test_split(data, labels,\n",
    "    test_size=0.20, stratify=labels, random_state=42)\n",
    "\n",
    "# initialize the training data augmentation object\n",
    "train_aug = ImageDataGenerator(\n",
    "    rotation_range=15,\n",
    "    fill_mode=\"nearest\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 5: load a pretrained network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Compiling model...\n"
     ]
    }
   ],
   "source": [
    "# load the VGG16 network, ensuring the head FC layer sets are left off\n",
    "base_model = VGG16(weights=\"imagenet\", include_top=False,\n",
    "\tinput_tensor=Input(shape=(224, 224, 3)))\n",
    "\n",
    "# construct the head of the model that will be placed on top of the the base model\n",
    "head_model = base_model.output\n",
    "head_model = AveragePooling2D(pool_size=(4, 4))(head_model)\n",
    "head_model = Flatten(name=\"flatten\")(head_model)\n",
    "head_model = Dense(64, activation=\"relu\")(head_model)\n",
    "head_model = Dropout(0.5)(head_model)\n",
    "head_model = Dense(2, activation=\"softmax\")(head_model)\n",
    "\n",
    "# place the head FC model on top of the base model (this will become\n",
    "# the actual model we will train)\n",
    "model = Model(inputs=base_model.input, outputs=head_model)\n",
    "\n",
    "# loop over all layers in the base model and freeze them so they will\n",
    "# *not* be updated during the first training process\n",
    "for layer in base_model.layers:\n",
    "    layer.trainable = False\n",
    "\n",
    "# compile our model\n",
    "print(\"Compiling model...\")\n",
    "opt = Adam(lr=INIT_LR, decay=INIT_LR / EPOCHS)\n",
    "\n",
    "model.compile(\n",
    "    loss=\"binary_crossentropy\", \n",
    "    optimizer=opt,\n",
    "\tmetrics=[\"accuracy\"]\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 6: train our model (this might take some time)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ttraining head...\n",
      "WARNING:tensorflow:sample_weight modes were coerced from\n",
      "  ...\n",
      "    to  \n",
      "  ['...']\n",
      "Train for 11 steps, validate on 23 samples\n",
      "Epoch 1/5\n",
      "11/11 [==============================] - 23s 2s/step - loss: 0.8243 - accuracy: 0.5679 - val_loss: 0.4560 - val_accuracy: 0.5625\n",
      "Epoch 2/5\n",
      "11/11 [==============================] - 23s 2s/step - loss: 0.6866 - accuracy: 0.6173 - val_loss: 0.4254 - val_accuracy: 0.7500\n",
      "Epoch 3/5\n",
      "11/11 [==============================] - 24s 2s/step - loss: 0.6111 - accuracy: 0.6173 - val_loss: 0.4153 - val_accuracy: 0.6875\n",
      "Epoch 4/5\n",
      "11/11 [==============================] - 24s 2s/step - loss: 0.6083 - accuracy: 0.6790 - val_loss: 0.4087 - val_accuracy: 0.6875\n",
      "Epoch 5/5\n",
      "11/11 [==============================] - 24s 2s/step - loss: 0.5746 - accuracy: 0.6543 - val_loss: 0.3701 - val_accuracy: 0.7500\n"
     ]
    }
   ],
   "source": [
    "# train the head of the network\n",
    "print(\"Ttraining head...\")\n",
    "history = model.fit(\n",
    "    train_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",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 7: evaluate the results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating network...\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "       covid       0.71      0.45      0.56        11\n",
      "      normal       0.62      0.83      0.71        12\n",
      "\n",
      "    accuracy                           0.65        23\n",
      "   macro avg       0.67      0.64      0.63        23\n",
      "weighted avg       0.67      0.65      0.64        23\n",
      "\n",
      "Confusion matrix:\n",
      " [[ 5  6]\n",
      " [ 2 10]]\n",
      "\n",
      "Accuracy: 0.6522\n",
      "Sensitivity: 0.4545\n",
      "Specificity: 0.8333\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# make predictions on the testing set\n",
    "print(\"Evaluating network...\")\n",
    "predictions = model.predict(testX, batch_size=BS)\n",
    "\n",
    "# for each image in the testing set we need to find the index of the\n",
    "# label with corresponding largest predicted probability\n",
    "predictions = np.argmax(predictions, axis=1)\n",
    "\n",
    "# show a nicely formatted classification report\n",
    "print(classification_report(testY.argmax(axis=1), predictions,\n",
    "\ttarget_names=lb.classes_))\n",
    "\n",
    "# compute the confusion matrix and and use it to derive the raw\n",
    "# accuracy, sensitivity, and specificity\n",
    "cm = confusion_matrix(testY.argmax(axis=1), predictions)\n",
    "total = sum(sum(cm))\n",
    "acc = (cm[0, 0] + cm[1, 1]) / total\n",
    "sensitivity = cm[0, 0] / (cm[0, 0] + cm[0, 1])\n",
    "specificity = cm[1, 1] / (cm[1, 0] + cm[1, 1])\n",
    "\n",
    "# show the confusion matrix, accuracy, sensitivity, and specificity\n",
    "print(\"Confusion matrix:\\n {}\\n\".format(cm))\n",
    "print(\"Accuracy: {:.4f}\".format(acc))\n",
    "print(\"Sensitivity: {:.4f}\".format(sensitivity))\n",
    "print(\"Specificity: {:.4f}\".format(specificity))\n",
    "\n",
    "# plot the training loss and accuracy\n",
    "xaxis = np.arange(1, EPOCHS+1)\n",
    "plt.style.use(\"ggplot\")\n",
    "plt.figure(figsize=(10,7))\n",
    "plt.plot(xaxis, history.history[\"loss\"], 'r--', label=\"train_loss\")\n",
    "plt.plot(xaxis, history.history[\"val_loss\"], 'b--', label=\"val_loss\")\n",
    "plt.plot(xaxis, history.history[\"accuracy\"], 'r', label=\"train_acc\")\n",
    "plt.plot(xaxis, history.history[\"val_accuracy\"], 'b', label=\"val_acc\")\n",
    "plt.title(\"Training Loss and Accuracy on COVID-19 Dataset\")\n",
    "plt.xlabel(\"Epoch #\")\n",
    "plt.ylabel(\"Loss/Accuracy\")\n",
    "plt.legend(loc=\"lower left\")\n",
    "plt.show()\n",
    "plt.savefig(plot_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving COVID-19 detector model...\n"
     ]
    }
   ],
   "source": [
    "# Save the model to disk\n",
    "print(\"Saving COVID-19 detector model...\")\n",
    "model.save(model_save_path, save_format=\"h5\")"
   ]
  }
 ],
 "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
}