24-CNN-cat_Vs_dog.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=\"http://akhavanpour.ir/notebook/images/srttu.gif\" alt=\"SRTTU\" style=\"width: 150px;\"/>\n",
    "\n",
    "[![Azure Notebooks](https://notebooks.azure.com/launch.png)](https://notebooks.azure.com/import/gh/Alireza-Akhavan/class.vision)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import keras\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">آموزش یک شبکه عصبی کانولوشنالی از ابتدا</div>\n",
    "\n",
    "<div style=\"direction:rtl;text-align:right;font-family:Tahoma\">\n",
    "کدها  برگرفته از فصل دو کتاب\n",
    "</div>\n",
    "\n",
    "[Deep Learning with Python](https://www.manning.com/books/deep-learning-with-python?a_aid=keras&a_bid=76564dff)\n",
    "<div style=\"direction:rtl;text-align:right;font-family:Tahoma\">\n",
    "و گیت هاب نویسنده کتاب و توسعه دهنده کراس \n",
    "</div>\n",
    "\n",
    "[François Chollet](http://nbviewer.jupyter.org/github/fchollet/deep-learning-with-python-notebooks/blob/master/5.2-using-convnets-with-small-datasets.ipynb)\n",
    "<div style=\"direction:rtl;text-align:right;font-family:Tahoma\">\n",
    "است.\n",
    "</div>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">مجموعه داده</div>\n",
    "\n",
    "<div style=\"direction:rtl;text-align:right;font-family:Tahoma\">\n",
    "مجموعه داده گربه در مقابل سگ برای این مثال استفاده شده است. این مجموعه داده در چالش بینایی کامپیوتر اواخر سال ۲۰۱۳ توسط سایت Kaggle.com در دسترس عموم قرار گرفت.\n",
    "<br>\n",
    "از قبل از سال ۲۰۱۳ استفاده از شبکه های کانولوشنالی خیلی رایج نبود.\n",
    "مجموعه داده در آدرس زیر قابل دانلود است.\n",
    "</div>\n",
    "\n",
    "`https://www.kaggle.com/c/dogs-vs-cats/data`\n",
    "\n",
    "\n",
    "\n",
    "![cats_vs_dogs_samples](https://s3.amazonaws.com/book.keras.io/img/ch5/cats_vs_dogs_samples.jpg)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Unsurprisingly, the cats vs. dogs Kaggle competition in 2013 was won by entrants who used convnets. The best entries could achieve up to \n",
    "95% accuracy. In our own example, we will get fairly close to this accuracy (in the next section), even though we will be training our \n",
    "models on less than 10% of the data that was available to the competitors.\n",
    "This original dataset contains 25,000 images of dogs and cats (12,500 from each class) and is 543MB large (compressed). After downloading \n",
    "and uncompressing it, we will create a new dataset containing three subsets: a training set with 1000 samples of each class, a validation \n",
    "set with 500 samples of each class, and finally a test set with 500 samples of each class.\n",
    "\n",
    "Here are a few lines of code to do this:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import os, shutil"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The path to the directory where the original\n",
    "# dataset was uncompressed\n",
    "original_dataset_dir = 'E:/dataset/catDog/train'\n",
    "\n",
    "# The directory where we will\n",
    "# store our smaller dataset\n",
    "base_dir = 'E:/dataset/catDog/catVsdog'\n",
    "os.mkdir(base_dir)\n",
    "\n",
    "# Directories for our training,\n",
    "# validation and test splits\n",
    "train_dir = os.path.join(base_dir, 'train')\n",
    "os.mkdir(train_dir)\n",
    "validation_dir = os.path.join(base_dir, 'validation')\n",
    "os.mkdir(validation_dir)\n",
    "test_dir = os.path.join(base_dir, 'test')\n",
    "os.mkdir(test_dir)\n",
    "\n",
    "# Directory with our training cat pictures\n",
    "train_cats_dir = os.path.join(train_dir, 'cats')\n",
    "os.mkdir(train_cats_dir)\n",
    "\n",
    "# Directory with our training dog pictures\n",
    "train_dogs_dir = os.path.join(train_dir, 'dogs')\n",
    "os.mkdir(train_dogs_dir)\n",
    "\n",
    "# Directory with our validation cat pictures\n",
    "validation_cats_dir = os.path.join(validation_dir, 'cats')\n",
    "os.mkdir(validation_cats_dir)\n",
    "\n",
    "# Directory with our validation dog pictures\n",
    "validation_dogs_dir = os.path.join(validation_dir, 'dogs')\n",
    "os.mkdir(validation_dogs_dir)\n",
    "\n",
    "# Directory with our validation cat pictures\n",
    "test_cats_dir = os.path.join(test_dir, 'cats')\n",
    "os.mkdir(test_cats_dir)\n",
    "\n",
    "# Directory with our validation dog pictures\n",
    "test_dogs_dir = os.path.join(test_dir, 'dogs')\n",
    "os.mkdir(test_dogs_dir)\n",
    "\n",
    "# Copy first 1000 cat images to train_cats_dir\n",
    "fnames = ['cat.{}.jpg'.format(i) for i in range(1000)]\n",
    "for fname in fnames:\n",
    "    src = os.path.join(original_dataset_dir, fname)\n",
    "    dst = os.path.join(train_cats_dir, fname)\n",
    "    shutil.copyfile(src, dst)\n",
    "\n",
    "# Copy next 500 cat images to validation_cats_dir\n",
    "fnames = ['cat.{}.jpg'.format(i) for i in range(1000, 1500)]\n",
    "for fname in fnames:\n",
    "    src = os.path.join(original_dataset_dir, fname)\n",
    "    dst = os.path.join(validation_cats_dir, fname)\n",
    "    shutil.copyfile(src, dst)\n",
    "    \n",
    "# Copy next 500 cat images to test_cats_dir\n",
    "fnames = ['cat.{}.jpg'.format(i) for i in range(1500, 2000)]\n",
    "for fname in fnames:\n",
    "    src = os.path.join(original_dataset_dir, fname)\n",
    "    dst = os.path.join(test_cats_dir, fname)\n",
    "    shutil.copyfile(src, dst)\n",
    "    \n",
    "# Copy first 1000 dog images to train_dogs_dir\n",
    "fnames = ['dog.{}.jpg'.format(i) for i in range(1000)]\n",
    "for fname in fnames:\n",
    "    src = os.path.join(original_dataset_dir, fname)\n",
    "    dst = os.path.join(train_dogs_dir, fname)\n",
    "    shutil.copyfile(src, dst)\n",
    "    \n",
    "# Copy next 500 dog images to validation_dogs_dir\n",
    "fnames = ['dog.{}.jpg'.format(i) for i in range(1000, 1500)]\n",
    "for fname in fnames:\n",
    "    src = os.path.join(original_dataset_dir, fname)\n",
    "    dst = os.path.join(validation_dogs_dir, fname)\n",
    "    shutil.copyfile(src, dst)\n",
    "    \n",
    "# Copy next 500 dog images to test_dogs_dir\n",
    "fnames = ['dog.{}.jpg'.format(i) for i in range(1500, 2000)]\n",
    "for fname in fnames:\n",
    "    src = os.path.join(original_dataset_dir, fname)\n",
    "    dst = os.path.join(test_dogs_dir, fname)\n",
    "    shutil.copyfile(src, dst)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\"> برای بررسی صحت انجام کار تعداد تصاویرآموزشی/آزمون/توسعه را بررسی میکنیم.</div>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total training cat images: 1000\n"
     ]
    }
   ],
   "source": [
    "print('total training cat images:', len(os.listdir(train_cats_dir)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total training dog images: 1000\n"
     ]
    }
   ],
   "source": [
    "print('total training dog images:', len(os.listdir(train_dogs_dir)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total validation cat images: 500\n"
     ]
    }
   ],
   "source": [
    "print('total validation cat images:', len(os.listdir(validation_cats_dir)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total validation dog images: 500\n"
     ]
    }
   ],
   "source": [
    "print('total validation dog images:', len(os.listdir(validation_dogs_dir)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total test cat images: 500\n"
     ]
    }
   ],
   "source": [
    "print('total test cat images:', len(os.listdir(test_cats_dir)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total test dog images: 500\n"
     ]
    }
   ],
   "source": [
    "print('total test dog images:', len(os.listdir(test_dogs_dir)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div style=\"direction:rtl;text-align:right;font-family:Tahoma\">\n",
    "با توجه به اینکه تعداد تصاویر یکسانی از سگ و گربه برداشته ایم معیار Accuracy معیارمناسبی برای ارزیابی خواهد بود.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">تعریف معماری مدل (model architecture)</div>\n",
    "\n",
    "Since we are attacking a binary classification problem, we are ending the network with a single unit (a `Dense` layer of size 1) and a \n",
    "`sigmoid` activation. This unit will encode the probability that the network is looking at one class or the other."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras import layers\n",
    "from keras import models\n",
    "\n",
    "model = models.Sequential()\n",
    "model.add(layers.Conv2D(32, (3, 3), activation='relu',\n",
    "                        input_shape=(150, 150, 3)))\n",
    "model.add(layers.MaxPooling2D((2, 2)))\n",
    "model.add(layers.Conv2D(64, (3, 3), activation='relu'))\n",
    "model.add(layers.MaxPooling2D((2, 2)))\n",
    "model.add(layers.Conv2D(128, (3, 3), activation='relu'))\n",
    "model.add(layers.MaxPooling2D((2, 2)))\n",
    "model.add(layers.Conv2D(128, (3, 3), activation='relu'))\n",
    "model.add(layers.MaxPooling2D((2, 2)))\n",
    "model.add(layers.Flatten())\n",
    "model.add(layers.Dense(512, activation='relu'))\n",
    "model.add(layers.Dense(1, activation='sigmoid'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at how the dimensions of the feature maps change with every successive layer:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 148, 148, 32)      896       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 74, 74, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 72, 72, 64)        18496     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 36, 36, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 34, 34, 128)       73856     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 17, 17, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 15, 15, 128)       147584    \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 7, 7, 128)         0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 6272)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 512)               3211776   \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 513       \n",
      "=================================================================\n",
      "Total params: 3,453,121\n",
      "Trainable params: 3,453,121\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For our compilation step, we'll go with the `RMSprop` optimizer as usual. Since we ended our network with a single sigmoid unit, we will \n",
    "use binary crossentropy as our loss (as a reminder, check out the table in Chapter 4, section 5 for a cheatsheet on what loss function to \n",
    "use in various situations)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras import optimizers\n",
    "\n",
    "model.compile(loss='binary_crossentropy',\n",
    "              optimizer=optimizers.RMSprop(lr=1e-4),\n",
    "              metrics=['acc'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">  پیش پردازش داده (Data preprocessing)</div>\n",
    "\n",
    "* Read the picture files.\n",
    "* Decode the JPEG content to RGB grids of pixels.\n",
    "* Convert these into floating point tensors.\n",
    "* Rescale the pixel values (between 0 and 255) to the [0, 1] interval (as you know, neural networks prefer to deal with small input values).\n",
    "\n",
    "It may seem a bit daunting, but thankfully Keras has utilities to take care of these steps automatically. Keras has a module with image \n",
    "processing helper tools, located at `keras.preprocessing.image`. In particular, it contains the class `ImageDataGenerator` which allows to \n",
    "quickly set up Python generators that can automatically turn image files on disk into batches of pre-processed tensors. This is what we \n",
    "will use here.\n",
    "\n",
    "<div style=\"direction:rtl;text-align:right;font-family:Tahoma\">\n",
    "    اطلاعات بیشتر در مستندات Keras :\n",
    "</div>\n",
    "\n",
    "`https://keras.io/preprocessing/image/`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2000 images belonging to 2 classes.\n",
      "Found 1000 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "# All images will be rescaled by 1./255\n",
    "train_datagen = ImageDataGenerator(rescale=1./255)\n",
    "test_datagen = ImageDataGenerator(rescale=1./255)\n",
    "\n",
    "train_generator = train_datagen.flow_from_directory(\n",
    "        # This is the target directory\n",
    "        train_dir,\n",
    "        # All images will be resized to 150x150\n",
    "        target_size=(150, 150),\n",
    "        batch_size=20,\n",
    "        # Since we use binary_crossentropy loss, we need binary labels\n",
    "        class_mode='binary')\n",
    "\n",
    "validation_generator = test_datagen.flow_from_directory(\n",
    "        validation_dir,\n",
    "        target_size=(150, 150),\n",
    "        batch_size=20,\n",
    "        class_mode='binary')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at the output of one of these generators: it yields batches of 150x150 RGB images (shape `(20, 150, 150, 3)`) and binary \n",
    "labels (shape `(20,)`). 20 is the number of samples in each batch (the batch size). Note that the generator yields these batches \n",
    "indefinitely: it just loops endlessly over the images present in the target folder. For this reason, we need to `break` the iteration loop \n",
    "at some point."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "data batch shape: (20, 150, 150, 3)\n",
      "labels batch shape: (20,)\n"
     ]
    }
   ],
   "source": [
    "for data_batch, labels_batch in train_generator:\n",
    "    print('data batch shape:', data_batch.shape)\n",
    "    print('labels batch shape:', labels_batch.shape)\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's fit our model to the data using the generator. We do it using the `fit_generator` method, the equivalent of `fit` for data generators \n",
    "like ours. It expects as first argument a Python generator that will yield batches of inputs and targets indefinitely, like ours does. \n",
    "Because the data is being generated endlessly, the generator needs to know example how many samples to draw from the generator before \n",
    "declaring an epoch over. This is the role of the `steps_per_epoch` argument: after having drawn `steps_per_epoch` batches from the \n",
    "generator, i.e. after having run for `steps_per_epoch` gradient descent steps, the fitting process will go to the next epoch. In our case, \n",
    "batches are 20-sample large, so it will take 100 batches until we see our target of 2000 samples.\n",
    "\n",
    "When using `fit_generator`, one may pass a `validation_data` argument, much like with the `fit` method. Importantly, this argument is \n",
    "allowed to be a data generator itself, but it could be a tuple of Numpy arrays as well. If you pass a generator as `validation_data`, then \n",
    "this generator is expected to yield batches of validation data endlessly, and thus you should also specify the `validation_steps` argument, \n",
    "which tells the process how many batches to draw from the validation generator for evaluation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "100/100 [==============================] - 10s 102ms/step - loss: 0.6871 - acc: 0.5430 - val_loss: 0.6706 - val_acc: 0.6100\n",
      "Epoch 2/30\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.6580 - acc: 0.6055 - val_loss: 0.6514 - val_acc: 0.6050\n",
      "Epoch 3/30\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.6075 - acc: 0.6705 - val_loss: 0.6088 - val_acc: 0.6720\n",
      "Epoch 4/30\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.5648 - acc: 0.7120 - val_loss: 0.5990 - val_acc: 0.6760\n",
      "Epoch 5/30\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.5386 - acc: 0.7240 - val_loss: 0.5730 - val_acc: 0.6970\n",
      "Epoch 6/30\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.5051 - acc: 0.7520 - val_loss: 0.5722 - val_acc: 0.6890\n",
      "Epoch 7/30\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.4834 - acc: 0.7710 - val_loss: 0.5501 - val_acc: 0.7100\n",
      "Epoch 8/30\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.4528 - acc: 0.7895 - val_loss: 0.5634 - val_acc: 0.7190\n",
      "Epoch 9/30\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.4284 - acc: 0.8035 - val_loss: 0.5988 - val_acc: 0.7000\n",
      "Epoch 10/30\n",
      "100/100 [==============================] - 7s 68ms/step - loss: 0.3983 - acc: 0.8230 - val_loss: 0.5738 - val_acc: 0.7190\n",
      "Epoch 11/30\n",
      "100/100 [==============================] - 7s 68ms/step - loss: 0.3780 - acc: 0.8310 - val_loss: 0.5705 - val_acc: 0.7150\n",
      "Epoch 12/30\n",
      "100/100 [==============================] - 7s 73ms/step - loss: 0.3460 - acc: 0.8575 - val_loss: 0.6450 - val_acc: 0.6910\n",
      "Epoch 13/30\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.3268 - acc: 0.8635 - val_loss: 0.5959 - val_acc: 0.7060\n",
      "Epoch 14/30\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.3011 - acc: 0.8665 - val_loss: 0.5675 - val_acc: 0.7310\n",
      "Epoch 15/30\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.2830 - acc: 0.8875 - val_loss: 0.6103 - val_acc: 0.7340\n",
      "Epoch 16/30\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.2567 - acc: 0.8945 - val_loss: 0.6139 - val_acc: 0.7270\n",
      "Epoch 17/30\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.2354 - acc: 0.9070 - val_loss: 0.6243 - val_acc: 0.7310\n",
      "Epoch 18/30\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.2176 - acc: 0.9215 - val_loss: 0.6302 - val_acc: 0.7340\n",
      "Epoch 19/30\n",
      "100/100 [==============================] - 7s 67ms/step - loss: 0.2015 - acc: 0.9300 - val_loss: 0.6495 - val_acc: 0.7320\n",
      "Epoch 20/30\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.1784 - acc: 0.9380 - val_loss: 0.7170 - val_acc: 0.7170\n",
      "Epoch 21/30\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.1616 - acc: 0.9365 - val_loss: 0.7690 - val_acc: 0.7180\n",
      "Epoch 22/30\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.1368 - acc: 0.9515 - val_loss: 0.7477 - val_acc: 0.7330\n",
      "Epoch 23/30\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.1200 - acc: 0.9620 - val_loss: 0.7869 - val_acc: 0.7210\n",
      "Epoch 24/30\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.1113 - acc: 0.9620 - val_loss: 0.7754 - val_acc: 0.7280\n",
      "Epoch 25/30\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0951 - acc: 0.9680 - val_loss: 0.9000 - val_acc: 0.7020\n",
      "Epoch 26/30\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0791 - acc: 0.9760 - val_loss: 0.8489 - val_acc: 0.7210\n",
      "Epoch 27/30\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0692 - acc: 0.9795 - val_loss: 0.8719 - val_acc: 0.7340\n",
      "Epoch 28/30\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0596 - acc: 0.9840 - val_loss: 0.9885 - val_acc: 0.7340\n",
      "Epoch 29/30\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0539 - acc: 0.9820 - val_loss: 0.9648 - val_acc: 0.7420\n",
      "Epoch 30/30\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0411 - acc: 0.9880 - val_loss: 1.1923 - val_acc: 0.7140\n"
     ]
    }
   ],
   "source": [
    "history = model.fit_generator(\n",
    "      train_generator,\n",
    "      steps_per_epoch=100,\n",
    "      epochs=30,\n",
    "      validation_data=validation_generator,\n",
    "      validation_steps=50)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's plot the loss and accuracy of the model over the training and validation data during training:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "These plots are characteristic of overfitting. Our training accuracy increases linearly over time, until it reaches nearly 100%, while our \n",
    "validation accuracy stalls at 70-72%. Our validation loss reaches its minimum after only five epochs then stalls, while the training loss \n",
    "keeps decreasing linearly until it reaches nearly 0.\n",
    "\n",
    "Because we only have relatively few training samples (2000), overfitting is going to be our number one concern.\n",
    "Let's train our network using data augmentation and dropout:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">شبکه با Dropout</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "100/100 [==============================] - 8s 76ms/step - loss: 0.6894 - acc: 0.5270 - val_loss: 0.6944 - val_acc: 0.5140\n",
      "Epoch 2/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.6607 - acc: 0.6160 - val_loss: 0.7345 - val_acc: 0.5090\n",
      "Epoch 3/100\n",
      "100/100 [==============================] - 7s 73ms/step - loss: 0.6228 - acc: 0.6675 - val_loss: 0.6389 - val_acc: 0.6400\n",
      "Epoch 4/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.5889 - acc: 0.6920 - val_loss: 0.6212 - val_acc: 0.6430\n",
      "Epoch 5/100\n",
      "100/100 [==============================] - 7s 73ms/step - loss: 0.5648 - acc: 0.7020 - val_loss: 0.7128 - val_acc: 0.6060\n",
      "Epoch 6/100\n",
      "100/100 [==============================] - 7s 65ms/step - loss: 0.5428 - acc: 0.7225 - val_loss: 0.5864 - val_acc: 0.6840\n",
      "Epoch 7/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.5207 - acc: 0.7430 - val_loss: 0.5631 - val_acc: 0.7090\n",
      "Epoch 8/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.5000 - acc: 0.7625 - val_loss: 0.5413 - val_acc: 0.7160\n",
      "Epoch 9/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.4690 - acc: 0.7725 - val_loss: 0.5477 - val_acc: 0.7170\n",
      "Epoch 10/100\n",
      "100/100 [==============================] - 7s 73ms/step - loss: 0.4568 - acc: 0.7830 - val_loss: 0.5365 - val_acc: 0.7260\n",
      "Epoch 11/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.4356 - acc: 0.7880 - val_loss: 0.5359 - val_acc: 0.7310\n",
      "Epoch 12/100\n",
      "100/100 [==============================] - 8s 76ms/step - loss: 0.4102 - acc: 0.8115 - val_loss: 0.5324 - val_acc: 0.7340\n",
      "Epoch 13/100\n",
      "100/100 [==============================] - 7s 73ms/step - loss: 0.4182 - acc: 0.8090 - val_loss: 0.6073 - val_acc: 0.7150\n",
      "Epoch 14/100\n",
      "100/100 [==============================] - 7s 74ms/step - loss: 0.3874 - acc: 0.8295 - val_loss: 0.5400 - val_acc: 0.7290\n",
      "Epoch 15/100\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.3835 - acc: 0.8280 - val_loss: 0.5678 - val_acc: 0.7240\n",
      "Epoch 16/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.3553 - acc: 0.8455 - val_loss: 0.5252 - val_acc: 0.7450\n",
      "Epoch 17/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.3465 - acc: 0.8460 - val_loss: 0.6187 - val_acc: 0.7160\n",
      "Epoch 18/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.3232 - acc: 0.8640 - val_loss: 0.5143 - val_acc: 0.7550\n",
      "Epoch 19/100\n",
      "100/100 [==============================] - 7s 74ms/step - loss: 0.3166 - acc: 0.8670 - val_loss: 0.5109 - val_acc: 0.7500\n",
      "Epoch 20/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.2992 - acc: 0.8735 - val_loss: 0.5157 - val_acc: 0.7600\n",
      "Epoch 21/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.2959 - acc: 0.8760 - val_loss: 0.6084 - val_acc: 0.7410\n",
      "Epoch 22/100\n",
      "100/100 [==============================] - 6s 64ms/step - loss: 0.2766 - acc: 0.8880 - val_loss: 0.5935 - val_acc: 0.7410\n",
      "Epoch 23/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.2722 - acc: 0.8820 - val_loss: 0.5240 - val_acc: 0.7600\n",
      "Epoch 24/100\n",
      "100/100 [==============================] - 8s 76ms/step - loss: 0.2446 - acc: 0.9040 - val_loss: 0.6517 - val_acc: 0.7340\n",
      "Epoch 25/100\n",
      "100/100 [==============================] - 7s 74ms/step - loss: 0.2330 - acc: 0.9130 - val_loss: 0.5708 - val_acc: 0.7520\n",
      "Epoch 26/100\n",
      "100/100 [==============================] - 8s 75ms/step - loss: 0.2248 - acc: 0.9070 - val_loss: 0.5257 - val_acc: 0.7700\n",
      "Epoch 27/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.2201 - acc: 0.9150 - val_loss: 0.5735 - val_acc: 0.7620\n",
      "Epoch 28/100\n",
      "100/100 [==============================] - 7s 67ms/step - loss: 0.2094 - acc: 0.9190 - val_loss: 0.5637 - val_acc: 0.7540\n",
      "Epoch 29/100\n",
      "100/100 [==============================] - 8s 76ms/step - loss: 0.2076 - acc: 0.9205 - val_loss: 0.5866 - val_acc: 0.7610\n",
      "Epoch 30/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.1827 - acc: 0.9340 - val_loss: 0.6123 - val_acc: 0.7560\n",
      "Epoch 31/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.1817 - acc: 0.9280 - val_loss: 0.6017 - val_acc: 0.7560\n",
      "Epoch 32/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.1704 - acc: 0.9315 - val_loss: 0.5948 - val_acc: 0.7600\n",
      "Epoch 33/100\n",
      "100/100 [==============================] - 8s 75ms/step - loss: 0.1663 - acc: 0.9355 - val_loss: 0.6496 - val_acc: 0.7560\n",
      "Epoch 34/100\n",
      "100/100 [==============================] - 7s 66ms/step - loss: 0.1591 - acc: 0.9370 - val_loss: 0.6076 - val_acc: 0.7620\n",
      "Epoch 35/100\n",
      "100/100 [==============================] - 7s 74ms/step - loss: 0.1461 - acc: 0.9485 - val_loss: 0.6325 - val_acc: 0.7640\n",
      "Epoch 36/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.1459 - acc: 0.9440 - val_loss: 0.5963 - val_acc: 0.7760\n",
      "Epoch 37/100\n",
      "100/100 [==============================] - 7s 73ms/step - loss: 0.1234 - acc: 0.9535 - val_loss: 0.7266 - val_acc: 0.7470\n",
      "Epoch 38/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.1391 - acc: 0.9445 - val_loss: 0.6515 - val_acc: 0.7570\n",
      "Epoch 39/100\n",
      "100/100 [==============================] - 8s 76ms/step - loss: 0.1280 - acc: 0.9570 - val_loss: 0.6692 - val_acc: 0.7570\n",
      "Epoch 40/100\n",
      "100/100 [==============================] - 7s 73ms/step - loss: 0.1138 - acc: 0.9600 - val_loss: 0.6916 - val_acc: 0.7650\n",
      "Epoch 41/100\n",
      "100/100 [==============================] - 7s 74ms/step - loss: 0.1084 - acc: 0.9645 - val_loss: 0.6601 - val_acc: 0.7640\n",
      "Epoch 42/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.1071 - acc: 0.9630 - val_loss: 0.7389 - val_acc: 0.7470\n",
      "Epoch 43/100\n",
      "100/100 [==============================] - 7s 73ms/step - loss: 0.1084 - acc: 0.9570 - val_loss: 0.6990 - val_acc: 0.7650\n",
      "Epoch 44/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.1045 - acc: 0.9660 - val_loss: 0.7916 - val_acc: 0.7490\n",
      "Epoch 45/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0898 - acc: 0.9685 - val_loss: 0.7113 - val_acc: 0.7720\n",
      "Epoch 46/100\n",
      "100/100 [==============================] - 7s 74ms/step - loss: 0.0993 - acc: 0.9620 - val_loss: 0.7143 - val_acc: 0.7710\n",
      "Epoch 47/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0807 - acc: 0.9720 - val_loss: 0.7336 - val_acc: 0.7670\n",
      "Epoch 48/100\n",
      "100/100 [==============================] - 7s 73ms/step - loss: 0.0923 - acc: 0.9670 - val_loss: 0.7492 - val_acc: 0.7540\n",
      "Epoch 49/100\n",
      "100/100 [==============================] - 8s 76ms/step - loss: 0.0791 - acc: 0.9725 - val_loss: 0.8292 - val_acc: 0.7680\n",
      "Epoch 50/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0799 - acc: 0.9720 - val_loss: 0.8038 - val_acc: 0.7630\n",
      "Epoch 51/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0786 - acc: 0.9715 - val_loss: 0.7994 - val_acc: 0.7650\n",
      "Epoch 52/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0595 - acc: 0.9805 - val_loss: 0.8475 - val_acc: 0.7630\n",
      "Epoch 53/100\n",
      "100/100 [==============================] - 7s 74ms/step - loss: 0.0658 - acc: 0.9770 - val_loss: 0.8780 - val_acc: 0.7530\n",
      "Epoch 54/100\n",
      "100/100 [==============================] - 8s 75ms/step - loss: 0.0624 - acc: 0.9765 - val_loss: 0.7878 - val_acc: 0.7580\n",
      "Epoch 55/100\n",
      "100/100 [==============================] - 7s 73ms/step - loss: 0.0535 - acc: 0.9815 - val_loss: 0.8381 - val_acc: 0.7610\n",
      "Epoch 56/100\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.0596 - acc: 0.9780 - val_loss: 0.8023 - val_acc: 0.7680\n",
      "Epoch 57/100\n",
      "100/100 [==============================] - 8s 75ms/step - loss: 0.0631 - acc: 0.9805 - val_loss: 0.8999 - val_acc: 0.7570\n",
      "Epoch 58/100\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.0535 - acc: 0.9830 - val_loss: 0.9586 - val_acc: 0.7510\n",
      "Epoch 59/100\n",
      "100/100 [==============================] - 6s 65ms/step - loss: 0.0523 - acc: 0.9800 - val_loss: 0.9511 - val_acc: 0.7570\n",
      "Epoch 60/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0507 - acc: 0.9810 - val_loss: 1.0322 - val_acc: 0.7550\n",
      "Epoch 61/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0542 - acc: 0.9800 - val_loss: 0.8776 - val_acc: 0.7590\n",
      "Epoch 62/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0547 - acc: 0.9810 - val_loss: 0.8419 - val_acc: 0.7670\n",
      "Epoch 63/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0485 - acc: 0.9810 - val_loss: 0.8661 - val_acc: 0.7730\n",
      "Epoch 64/100\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.0427 - acc: 0.9865 - val_loss: 0.9159 - val_acc: 0.7590\n",
      "Epoch 65/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0567 - acc: 0.9780 - val_loss: 0.9303 - val_acc: 0.7580\n",
      "Epoch 66/100\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.0465 - acc: 0.9835 - val_loss: 0.9343 - val_acc: 0.7530\n",
      "Epoch 67/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0407 - acc: 0.9840 - val_loss: 0.9852 - val_acc: 0.7570\n",
      "Epoch 68/100\n",
      "100/100 [==============================] - 7s 74ms/step - loss: 0.0455 - acc: 0.9840 - val_loss: 0.9251 - val_acc: 0.7600\n",
      "Epoch 69/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0358 - acc: 0.9890 - val_loss: 1.0259 - val_acc: 0.7530\n",
      "Epoch 70/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0392 - acc: 0.9865 - val_loss: 1.0112 - val_acc: 0.7540\n",
      "Epoch 71/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0382 - acc: 0.9855 - val_loss: 1.0240 - val_acc: 0.7520\n",
      "Epoch 72/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0456 - acc: 0.9825 - val_loss: 1.0001 - val_acc: 0.7630\n",
      "Epoch 73/100\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.0364 - acc: 0.9860 - val_loss: 1.0619 - val_acc: 0.7650\n",
      "Epoch 74/100\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.0367 - acc: 0.9850 - val_loss: 1.0581 - val_acc: 0.7650\n",
      "Epoch 75/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0267 - acc: 0.9895 - val_loss: 0.9538 - val_acc: 0.7720\n",
      "Epoch 76/100\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.0366 - acc: 0.9900 - val_loss: 1.0046 - val_acc: 0.7780\n",
      "Epoch 77/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0402 - acc: 0.9850 - val_loss: 0.9360 - val_acc: 0.7790\n",
      "Epoch 78/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0352 - acc: 0.9900 - val_loss: 1.0473 - val_acc: 0.7550\n",
      "Epoch 79/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0399 - acc: 0.9865 - val_loss: 1.0465 - val_acc: 0.7650\n",
      "Epoch 80/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0224 - acc: 0.9925 - val_loss: 1.0899 - val_acc: 0.7620\n",
      "Epoch 81/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0364 - acc: 0.9860 - val_loss: 0.9956 - val_acc: 0.7590\n",
      "Epoch 82/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0263 - acc: 0.9930 - val_loss: 1.0581 - val_acc: 0.7750\n",
      "Epoch 83/100\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.0262 - acc: 0.9900 - val_loss: 0.9802 - val_acc: 0.7780\n",
      "Epoch 84/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0342 - acc: 0.9885 - val_loss: 1.0599 - val_acc: 0.7700\n",
      "Epoch 85/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0252 - acc: 0.9915 - val_loss: 1.0681 - val_acc: 0.7780\n",
      "Epoch 86/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0292 - acc: 0.9915 - val_loss: 1.1017 - val_acc: 0.7670\n",
      "Epoch 87/100\n",
      "100/100 [==============================] - 7s 72ms/step - loss: 0.0336 - acc: 0.9885 - val_loss: 1.0666 - val_acc: 0.7690\n",
      "Epoch 88/100\n",
      "100/100 [==============================] - 8s 76ms/step - loss: 0.0289 - acc: 0.9880 - val_loss: 1.1479 - val_acc: 0.7730\n",
      "Epoch 89/100\n",
      "100/100 [==============================] - 8s 75ms/step - loss: 0.0330 - acc: 0.9890 - val_loss: 1.1886 - val_acc: 0.7650\n",
      "Epoch 90/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0305 - acc: 0.9890 - val_loss: 1.1070 - val_acc: 0.7640\n",
      "Epoch 91/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0294 - acc: 0.9895 - val_loss: 1.0466 - val_acc: 0.7720\n",
      "Epoch 92/100\n",
      "100/100 [==============================] - 7s 74ms/step - loss: 0.0219 - acc: 0.9930 - val_loss: 1.0923 - val_acc: 0.7710\n",
      "Epoch 93/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0264 - acc: 0.9905 - val_loss: 1.2646 - val_acc: 0.7570\n",
      "Epoch 94/100\n",
      "100/100 [==============================] - 6s 65ms/step - loss: 0.0272 - acc: 0.9890 - val_loss: 1.1050 - val_acc: 0.7730\n",
      "Epoch 95/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0282 - acc: 0.9905 - val_loss: 1.1346 - val_acc: 0.7680\n",
      "Epoch 96/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0249 - acc: 0.9900 - val_loss: 1.0549 - val_acc: 0.7660\n",
      "Epoch 97/100\n",
      "100/100 [==============================] - 7s 67ms/step - loss: 0.0256 - acc: 0.9910 - val_loss: 1.2847 - val_acc: 0.7550\n",
      "Epoch 98/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 0.0198 - acc: 0.9945 - val_loss: 1.1909 - val_acc: 0.7830\n",
      "Epoch 99/100\n",
      "100/100 [==============================] - 7s 69ms/step - loss: 0.0320 - acc: 0.9900 - val_loss: 1.1163 - val_acc: 0.7730\n",
      "Epoch 100/100\n",
      "100/100 [==============================] - 7s 71ms/step - loss: 0.0219 - acc: 0.9925 - val_loss: 1.4962 - val_acc: 0.7470\n"
     ]
    }
   ],
   "source": [
    "model = models.Sequential()\n",
    "model.add(layers.Conv2D(32, (3, 3), activation='relu',\n",
    "                        input_shape=(150, 150, 3)))\n",
    "model.add(layers.MaxPooling2D((2, 2)))\n",
    "model.add(layers.Conv2D(64, (3, 3), activation='relu'))\n",
    "model.add(layers.MaxPooling2D((2, 2)))\n",
    "model.add(layers.Conv2D(128, (3, 3), activation='relu'))\n",
    "model.add(layers.MaxPooling2D((2, 2)))\n",
    "model.add(layers.Conv2D(128, (3, 3), activation='relu'))\n",
    "model.add(layers.MaxPooling2D((2, 2)))\n",
    "model.add(layers.Flatten())\n",
    "model.add(layers.Dropout(0.5))\n",
    "model.add(layers.Dense(512, activation='relu'))\n",
    "model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "model.compile(loss='binary_crossentropy',\n",
    "              optimizer=optimizers.RMSprop(lr=1e-4),\n",
    "              metrics=['acc'])\n",
    "\n",
    "history = model.fit_generator(\n",
    "      train_generator,\n",
    "      steps_per_epoch=100,\n",
    "      epochs=100,\n",
    "      validation_data=validation_generator,\n",
    "      validation_steps=50)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's plot our results again:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">ذخیره کردن مدل</div>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.save('cats_and_dogs_small_1.h5')"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda env:virtual_platform]",
   "language": "python",
   "name": "conda-env-virtual_platform-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}