Untitled-1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "\n",
    "from tensorflow.keras import optimizers, regularizers\n",
    "from tensorflow.keras.utils import load_img, img_to_array\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Flatten, Dropout, BatchNormalization, Conv2D, MaxPooling2D\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import cv2\n",
    "import os\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 30250 images belonging to 2 classes.\n",
      "Found 6300 images belonging to 2 classes.\n",
      "Found 6300 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "train_path = 'data/train'\n",
    "test_path = 'data/test'\n",
    "valid_path = 'data/valid'\n",
    "\n",
    "image_shape = (350,350,3)\n",
    "N_CLASSES = 2\n",
    "BATCH_SIZE = 256\n",
    "\n",
    "# loading training data and rescaling it using ImageDataGenerator\n",
    "train_datagen = ImageDataGenerator(dtype='float32', rescale= 1./255.)\n",
    "train_generator = train_datagen.flow_from_directory(train_path,\n",
    "                                                   batch_size = BATCH_SIZE,\n",
    "                                                   target_size = (350,350),\n",
    "                                                   class_mode = 'categorical',\n",
    "                                                   shuffle = True)\n",
    "\n",
    "# loading validation data and rescaling it using ImageDataGenerator\n",
    "valid_datagen = ImageDataGenerator(dtype='float32', rescale= 1./255.)\n",
    "valid_generator = valid_datagen.flow_from_directory(valid_path,\n",
    "                                                   batch_size = BATCH_SIZE,\n",
    "                                                   target_size = (350,350),\n",
    "                                                   class_mode = 'categorical',\n",
    "                                                   shuffle = True)\n",
    "\n",
    "# loading test data and rescaling it using ImageDataGenerator\n",
    "test_datagen = ImageDataGenerator(dtype='float32', rescale = 1.0/255.0)\n",
    "test_generator = test_datagen.flow_from_directory(test_path,\n",
    "                                                   batch_size = BATCH_SIZE,\n",
    "                                                   target_size = (350,350),\n",
    "                                                   class_mode = 'categorical',\n",
    "                                                   shuffle = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d (Conv2D)             (None, 349, 349, 8)       104       \n",
      "                                                                 \n",
      " max_pooling2d (MaxPooling2D  (None, 174, 174, 8)      0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " conv2d_1 (Conv2D)           (None, 173, 173, 16)      528       \n",
      "                                                                 \n",
      " max_pooling2d_1 (MaxPooling  (None, 86, 86, 16)       0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " conv2d_2 (Conv2D)           (None, 85, 85, 32)        2080      \n",
      "                                                                 \n",
      " max_pooling2d_2 (MaxPooling  (None, 42, 42, 32)       0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " dropout (Dropout)           (None, 42, 42, 32)        0         \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 56448)             0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 300)               16934700  \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 300)               0         \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 2)                 602       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 16,938,014\n",
      "Trainable params: 16,938,014\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# defining the coefficient that our regularizer will use\n",
    "weight_decay = 1e-3\n",
    "\n",
    "# building a sequential CNN model and adding layers to it\n",
    "# dropout and the regularizer are used in general to prevent overfitting\n",
    "first_model = Sequential([\n",
    "    Conv2D(filters = 8 , kernel_size = 2, activation = 'relu', \n",
    "    input_shape = image_shape), MaxPooling2D(pool_size = 2),\n",
    "    \n",
    "    Conv2D(filters = 16 , kernel_size = 2, activation = 'relu', \n",
    "    input_shape = image_shape), MaxPooling2D(pool_size = 2),\n",
    "    \n",
    "    Conv2D(filters = 32 , kernel_size = 2, activation = 'relu',\n",
    "           kernel_regularizer = regularizers.l2(weight_decay)),\n",
    "    MaxPooling2D(pool_size = 2),\n",
    "    \n",
    "    Dropout(0.4),\n",
    "    Flatten(),\n",
    "    Dense(300,activation='relu'),\n",
    "    Dropout(0.5),\n",
    "    Dense(2,activation='softmax')\n",
    "])\n",
    "# showing the summary of our model (layers and number of parameters)\n",
    "first_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-08-31 09:45:19.936112: W tensorflow/core/platform/profile_utils/cpu_utils.cc:128] Failed to get CPU frequency: 0 Hz\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "119/119 [==============================] - ETA: 0s - loss: 0.6009 - auc: 0.8003 - acc: 0.6815\n",
      "Epoch 1: val_loss improved from inf to 0.45137, saving model to first_model.hdf5\n",
      "119/119 [==============================] - 135s 1s/step - loss: 0.6009 - auc: 0.8003 - acc: 0.6815 - val_loss: 0.4514 - val_auc: 0.9409 - val_acc: 0.8805\n",
      "Epoch 2/5\n",
      "119/119 [==============================] - ETA: 0s - loss: 0.3822 - auc: 0.9404 - acc: 0.8801\n",
      "Epoch 2: val_loss improved from 0.45137 to 0.30627, saving model to first_model.hdf5\n",
      "119/119 [==============================] - 134s 1s/step - loss: 0.3822 - auc: 0.9404 - acc: 0.8801 - val_loss: 0.3063 - val_auc: 0.9622 - val_acc: 0.9016\n",
      "Epoch 3/5\n",
      "119/119 [==============================] - ETA: 0s - loss: 0.3089 - auc: 0.9532 - acc: 0.8907\n",
      "Epoch 3: val_loss improved from 0.30627 to 0.26763, saving model to first_model.hdf5\n",
      "119/119 [==============================] - 135s 1s/step - loss: 0.3089 - auc: 0.9532 - acc: 0.8907 - val_loss: 0.2676 - val_auc: 0.9671 - val_acc: 0.9035\n",
      "Epoch 4/5\n",
      "119/119 [==============================] - ETA: 0s - loss: 0.2867 - auc: 0.9579 - acc: 0.8953\n",
      "Epoch 4: val_loss improved from 0.26763 to 0.25326, saving model to first_model.hdf5\n",
      "119/119 [==============================] - 136s 1s/step - loss: 0.2867 - auc: 0.9579 - acc: 0.8953 - val_loss: 0.2533 - val_auc: 0.9696 - val_acc: 0.9071\n",
      "Epoch 5/5\n",
      "119/119 [==============================] - ETA: 0s - loss: 0.2756 - auc: 0.9610 - acc: 0.8984\n",
      "Epoch 5: val_loss did not improve from 0.25326\n",
      "119/119 [==============================] - 137s 1s/step - loss: 0.2756 - auc: 0.9610 - acc: 0.8984 - val_loss: 0.2585 - val_auc: 0.9671 - val_acc: 0.9063\n"
     ]
    }
   ],
   "source": [
    "# don't stop everything if an image didn't load correctly\n",
    "from PIL import ImageFile\n",
    "ImageFile.LOAD_TRUNCATED_IMAGES = True\n",
    "\n",
    "# checkpointer to save the model only if it improved\n",
    "checkpointer = ModelCheckpoint('first_model.hdf5',verbose=1, save_best_only= True)\n",
    "# early stopping to stop the training if our validation loss didn't decrease for (10) consecutive epochs\n",
    "early_stopping = EarlyStopping(monitor= 'val_loss', patience= 10)\n",
    "# Adam, best optimiser for deep learning models to help with the training\n",
    "optimizer = optimizers.Adam(learning_rate= 0.00001, decay= 1e-5)\n",
    "# setting our loss function and which metric to evaluate\n",
    "first_model.compile(loss= 'categorical_crossentropy', optimizer= optimizer,\n",
    "                    metrics=['AUC','acc'])\n",
    "\n",
    "# TRAIN\n",
    "history = first_model.fit(train_generator,\n",
    "                    epochs = 5,\n",
    "                    verbose = 1,\n",
    "                    validation_data = valid_generator,\n",
    "                    callbacks = [checkpointer, early_stopping])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# add history of accuracy and validation accuracy to the plot\n",
    "plt.plot(history.history['acc'], label = 'train',)\n",
    "plt.plot(history.history['val_acc'], label = 'valid')\n",
    "\n",
    "# adding legend and labels\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('accuracy')\n",
    "\n",
    "# show the plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "25/25 [==============================] - 12s 450ms/step - loss: 0.2543 - auc: 0.9683 - acc: 0.9159\n"
     ]
    }
   ],
   "source": [
    "# see if it's good at predecting new inputs\n",
    "result = first_model.evaluate(test_generator)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "25/25 [==============================] - 12s 456ms/step\n"
     ]
    }
   ],
   "source": [
    "# see if it's good at predecting new inputs\n",
    "result = first_model.predict(test_generator)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.9656742 , 0.03432578],\n",
       "       [0.9956352 , 0.00436479],\n",
       "       [0.9791196 , 0.02088046],\n",
       "       ...,\n",
       "       [0.33177572, 0.6682243 ],\n",
       "       [0.06591696, 0.934083  ],\n",
       "       [0.04055419, 0.9594458 ]], dtype=float32)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x2d3be3e80>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "class_labels = list(test_generator.class_indices.keys())\n",
    "y_prediction = result.argmax(axis=1)\n",
    "\n",
    "# Sklearn\n",
    "from sklearn.metrics import accuracy_score, confusion_matrix, ConfusionMatrixDisplay\n",
    "conf_matrix = confusion_matrix(y_true=test_generator.classes, y_pred=y_prediction)\n",
    "ConfusionMatrixDisplay(conf_matrix, display_labels=class_labels).plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "25/25 [==============================] - 11s 447ms/step\n"
     ]
    }
   ],
   "source": [
    "predicted_probabilities = first_model.predict(test_generator)\n",
    "predicted_labels = np.argmax(predicted_probabilities, axis=1)\n",
    "\n",
    "ground_truth_labels = test_generator.classes\n",
    "\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "cm = confusion_matrix(ground_truth_labels, predicted_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[2607,  213],\n",
       "       [ 317, 3163]])"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cm"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "CAS_courses_env",
   "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": "3.10.8"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}