transfer_learning_VGG19.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Classification of Anomalies in Gastrointestinal Tract through Endoscopic Imagery with Deep Learning\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### we used opencv for preproccesing the image dataset\n",
    "### we used tensorflow and keras libarary for machine learning stuff "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "import cv2\n",
    "import matplotlib.pyplot as plt\n",
    "import os\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import tensorflowjs as tfjs\n",
    "from tensorflow.keras import applications\n",
    "from tensorflow.keras import optimizers\n",
    "import argparse\n",
    "from imutils import paths\n",
    "import np_utils"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### first we set the image dataset directory and define the categories "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "datadir='/home/shashimal/Documents/dataset'\n",
    "categories = ['dyed-lifted-polyps','dyed-resection-margins','esophagitis',\n",
    "             'normal-cecum','normal-pylorus','normal-z-line','polyps','ulcerative-colitis']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### using the opencv, we read images from dataset directory and resize images to 96*96\n",
    "### then insert the resized image with class lable to array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "training_data=[]\n",
    "def create_training_data():\n",
    "    for category in categories:\n",
    "        path = os.path.join(datadir,category)\n",
    "        class_num = categories.index(category)\n",
    "        for img in os.listdir(path):\n",
    "            try:\n",
    "                img_array = cv2.imread(os.path.join(path,img))\n",
    "                new_array = cv2.resize(img_array,(96,96))\n",
    "#                crop_array = new_array[0:96,35:96]\n",
    "                training_data.append([new_array,class_num])\n",
    "            except Exception as e:    \n",
    "                pass\n",
    "create_training_data() "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###  append features and labels to separated list\n",
    "### then convert the feature array into a NumPy array and reshape the array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8000, 96, 96, 3)"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x=[]\n",
    "y=[]\n",
    "\n",
    "for features,label in training_data:\n",
    "    x.append(features)\n",
    "    y.append(label)\n",
    "x = np.array(x).reshape(8000,96,96,3)\n",
    "x.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### partition the data into training and testing splits using 80% of the data for training and the remaining 20% for testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "x_train,x_test,y_train,y_test = train_test_split(x,y,test_size=0.2,random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### then preprocess dataset by scaling all pixel intensities to the range [0, 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train=x_train.astype('float32')\n",
    "x_test=x_test.astype('float32')\n",
    "x_train/=255\n",
    "x_test/=255"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### encode the labels (which are currently strings) as integers and then one-hot encode them"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train = tf.keras.utils.to_categorical(y_train, 8)\n",
    "y_test = tf.keras.utils.to_categorical(y_test, 8)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### import VGG19 model (for Transfer learning)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = applications.VGG19(weights = \"imagenet\", include_top=False, input_shape = (96, 96, 3))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### first we iterate through the model and make 20 layer non-trainable\n",
    "### then add pooling layer and add some dense layers with relu activation\n",
    "### we used categorical_crossentropy loss function and Adam as the optimizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "for layer in model.layers[:20]:\n",
    "    layer.trainable = False\n",
    "x = model.output\n",
    "x =  tf.keras.layers.GlobalAveragePooling2D()(x)\n",
    "x = tf.keras.layers.Dense(1024, activation=\"relu\")(x)\n",
    "x = tf.keras.layers.Dropout(0.5)(x)\n",
    "x = tf.keras.layers.Dense(512, activation=\"relu\")(x)\n",
    "x = tf.keras.layers.Dropout(0.5)(x)\n",
    "x = tf.keras.layers.Dense(256, activation=\"relu\")(x)\n",
    "x = tf.keras.layers.Dropout(0.5)(x)\n",
    "x = tf.keras.layers.Dense(128, activation=\"relu\")(x)\n",
    "predictions = tf.keras.layers.Dense(8, activation=\"softmax\")(x)   \n",
    "model_final = tf.keras.models.Model(model.input,predictions)\n",
    "model_final.compile(loss=\"categorical_crossentropy\", optimizer='Adam', metrics=[\"accuracy\"])\n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### then we fit the dataset to train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 6400 samples, validate on 1600 samples\n",
      "Epoch 1/8\n",
      "6400/6400 [==============================] - 542s 85ms/sample - loss: 0.7060 - acc: 0.6850 - val_loss: 0.5868 - val_acc: 0.7563\n",
      "Epoch 2/8\n",
      "6400/6400 [==============================] - 533s 83ms/sample - loss: 0.5851 - acc: 0.7450 - val_loss: 0.5606 - val_acc: 0.7444\n",
      "Epoch 3/8\n",
      "6400/6400 [==============================] - 531s 83ms/sample - loss: 0.5352 - acc: 0.7783 - val_loss: 0.5574 - val_acc: 0.7744\n",
      "Epoch 4/8\n",
      "6400/6400 [==============================] - 530s 83ms/sample - loss: 0.4642 - acc: 0.8094 - val_loss: 0.4981 - val_acc: 0.7981\n",
      "Epoch 5/8\n",
      "6400/6400 [==============================] - 497s 78ms/sample - loss: 0.4255 - acc: 0.8236 - val_loss: 0.5117 - val_acc: 0.8112\n",
      "Epoch 6/8\n",
      "6400/6400 [==============================] - 491s 77ms/sample - loss: 0.3843 - acc: 0.8414 - val_loss: 0.5850 - val_acc: 0.7912\n",
      "Epoch 7/8\n",
      "6400/6400 [==============================] - 502s 78ms/sample - loss: 0.3606 - acc: 0.8562 - val_loss: 0.5147 - val_acc: 0.8050\n",
      "Epoch 8/8\n",
      "6400/6400 [==============================] - 517s 81ms/sample - loss: 0.3165 - acc: 0.8714 - val_loss: 0.6057 - val_acc: 0.8037\n"
     ]
    }
   ],
   "source": [
    "history=model_final.fit(x_train, y_train,\n",
    "          batch_size=32,\n",
    "          epochs=8,\n",
    "          verbose=1,\n",
    "          validation_data=(x_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "score = model_final.evaluate(x_test, y_test, verbose=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.6057374227046967\n",
      "Test accuracy: 0.80375\n"
     ]
    }
   ],
   "source": [
    "print('Test loss:', score[0])\n",
    "print('Test accuracy:', score[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'acc', 'val_loss', 'val_acc'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "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": [
    "print(history.history.keys())\n",
    "\n",
    "plt.plot(history.history['acc'])\n",
    "plt.plot(history.history['val_acc'])\n",
    "plt.title('model accuracy')\n",
    "plt.ylabel('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.7406090e-01, 8.2465869e-01, 7.4679956e-05, 6.4581633e-05,\n",
       "       3.0395921e-04, 3.3695229e-05, 7.4848841e-04, 5.4951608e-05],\n",
       "      dtype=float32)"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0., 1., 0., 0., 0., 0., 0., 0.], dtype=float32)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}