5_CNN_Keras_TF.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Classification: CATS vs DOGS \n",
    "    jpeg -> CNN Keras -> [0,1]\n",
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Import Tensorflow with Keras and Packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "from time import time\n",
    "from tensorflow.python.keras.callbacks import TensorBoard"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Conv2D, MaxPooling2D, Activation, Dropout, Flatten\n",
    "from keras.utils import to_categorical"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import backend as K"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.preprocessing.image import array_to_img, img_to_array, load_img, ImageDataGenerator\n",
    "from PIL import Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "%matplotlib inline "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "### FOLDERS and CATEGORIES:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Folders"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "datadir = 'dogVScat'\n",
    "cat1 = 'dogs'\n",
    "cat2 = 'cats'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Training set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "target = os.chdir(datadir + \"/train/\" + cat1)\n",
    "cat_1 = os.listdir()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "target = os.chdir(\"../\" + cat2)\n",
    "cat_2 = os.listdir()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CAT1: 5501 CAT2: 6054 TOTAL: 11555\n"
     ]
    }
   ],
   "source": [
    "print(\"CAT1:\",len(cat_1),\"CAT2:\",len(cat_2), \"TOTAL:\",len(cat_1)+len(cat_2))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Validation set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "target = os.chdir(\"../../validation/\" + cat1)\n",
    "cat_1_val = os.listdir()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "target = os.chdir(\"../\" + cat2)\n",
    "cat_2_val = os.listdir()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CAT1: 1302 CAT2: 1500 TOTAL: 2802\n"
     ]
    }
   ],
   "source": [
    "print(\"CAT1:\",len(cat_1_val),\"CAT2:\",len(cat_2_val), \"TOTAL:\",len(cat_1_val)+len(cat_2_val))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Return to Home"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "os.chdir(\"../../../\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "### HYPER PARAMETERS:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Dimensions\n",
    "- Epochs\n",
    "- Batch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "###\n",
    "img_width, img_height = 300, 300\n",
    "\n",
    "data_dir = 'dogVScat'\n",
    "train_data_dir = data_dir + '/train'\n",
    "validation_data_dir = data_dir + '/validation'\n",
    "test_data_dir = data_dir + '/test'\n",
    "\n",
    "nb_train_samples = len(cat_1)+len(cat_2)\n",
    "nb_validation_samples = len(cat_1_val)+len(cat_2_val)\n",
    "epochs = 5                                                 # 50\n",
    "batch_size = 32                                            # 16"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "### CNN MODEL:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Layers Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From c:\\users\\e.petrachi\\appdata\\local\\continuum\\anaconda3\\envs\\tf-env\\lib\\site-packages\\tensorflow\\python\\framework\\op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n",
      "WARNING:tensorflow:From c:\\users\\e.petrachi\\appdata\\local\\continuum\\anaconda3\\envs\\tf-env\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 298, 298, 32)      896       \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 298, 298, 32)      0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 149, 149, 32)      0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 147, 147, 32)      9248      \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 147, 147, 32)      0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 73, 73, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 71, 71, 64)        18496     \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 71, 71, 64)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 35, 35, 64)        0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 78400)             0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 64)                5017664   \n",
      "_________________________________________________________________\n",
      "activation_4 (Activation)    (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 65        \n",
      "_________________________________________________________________\n",
      "activation_5 (Activation)    (None, 1)                 0         \n",
      "=================================================================\n",
      "Total params: 5,046,369\n",
      "Trainable params: 5,046,369\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "if K.image_data_format() == 'channels_first':\n",
    "    input_shape = (3, img_width, img_height)\n",
    "else:\n",
    "    input_shape = (img_width, img_height, 3)\n",
    "\n",
    "model = Sequential()\n",
    "model.add(Conv2D(32, (3, 3), input_shape=input_shape))\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "\n",
    "model.add(Conv2D(32, (3, 3)))\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "\n",
    "model.add(Conv2D(64, (3, 3)))\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "\n",
    "model.add(Flatten())\n",
    "model.add(Dense(64))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(1))\n",
    "model.add(Activation('sigmoid'))\n",
    "\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss='binary_crossentropy', optimizer='rmsprop', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Data Generators"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_datagen = ImageDataGenerator(\n",
    "    rescale=1. / 255,\n",
    "    shear_range=0.2,\n",
    "    zoom_range=0.2,\n",
    "    horizontal_flip=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_datagen = ImageDataGenerator(rescale=1. / 255)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Train VS Validation SET"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 11555 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "train_generator = train_datagen.flow_from_directory(\n",
    "    train_data_dir,\n",
    "    target_size=(img_width, img_height),\n",
    "    batch_size=batch_size,\n",
    "    class_mode='binary')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2802 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "validation_generator = test_datagen.flow_from_directory(\n",
    "    validation_data_dir,\n",
    "    target_size=(img_width, img_height),\n",
    "    batch_size=batch_size,\n",
    "    class_mode='binary')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Visualize on Tensorboard "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "tensorboard = TensorBoard(log_dir=\"logs/{}\".format(time()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Model Training "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "361/361 [==============================] - 553s 2s/step - loss: 0.5096 - acc: 0.7584 - val_loss: 0.5269 - val_acc: 0.7845\n",
      "Epoch 2/5\n",
      "361/361 [==============================] - 564s 2s/step - loss: 0.5045 - acc: 0.7630 - val_loss: 0.4669 - val_acc: 0.7888\n",
      "Epoch 3/5\n",
      "361/361 [==============================] - 585s 2s/step - loss: 0.4921 - acc: 0.7645 - val_loss: 0.4461 - val_acc: 0.8036\n",
      "Epoch 4/5\n",
      "361/361 [==============================] - 544s 2s/step - loss: 0.4856 - acc: 0.7759 - val_loss: 0.4608 - val_acc: 0.7957\n",
      "Epoch 5/5\n",
      "361/361 [==============================] - 559s 2s/step - loss: 0.4821 - acc: 0.7832 - val_loss: 0.5391 - val_acc: 0.7996\n"
     ]
    }
   ],
   "source": [
    "history = model.fit_generator(\n",
    "    train_generator,\n",
    "    steps_per_epoch=nb_train_samples // batch_size,\n",
    "    epochs=epochs,\n",
    "    validation_data=validation_generator,\n",
    "    validation_steps=nb_validation_samples // batch_size, callbacks=[tensorboard])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Visualize Learning Rate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "plt.plot(history.history['acc'])\n",
    "plt.plot(history.history['val_acc'])\n",
    "\n",
    "plt.title('Model Accuracy')\n",
    "plt.ylabel('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['cnn:train', 'cnn:test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "\n",
    "plt.title('Model Loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['cnn:train', 'cnn:test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Save"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save_weights('keras_weights.h5')\n",
    "model.save('keras_model.h5') "
   ]
  },
  {
   "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}