08. Generating Images using GANs/8.07 Implementing DCGAN to Generate CIFAR Images.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Implementing DCGAN to Generate CIFAR Images"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, we will see how to implement DCGAN in tensorflow. We will learn how to use DCGAN to generate images from the CIFAR-10 dataset. CIFAR-10 (Canadian Institute For Advanced Research) consists of 60,000 images from 10 different classes which include airplanes, cars, birds, cats, deer, dogs, frogs, horses, ships, and trucks. We will examine how can we use DCGAN to generate such images. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Import libraries\n",
    "\n",
    "\n",
    "First, import the required libraries:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "tf.logging.set_verbosity(tf.logging.ERROR)\n",
    "\n",
    "from keras.datasets import cifar10\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "from IPython import display\n",
    "\n",
    "from scipy.misc import toimage"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Explore the Dataset\n",
    "\n",
    "Load the CIFAR dataset:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "(x_train, y_train), _ = cifar10.load_data()\n",
    "\n",
    "#normalize the data\n",
    "x_train = x_train.astype('float32')/255.0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let us see what we have in our dataset. Define a helper function for plotting the image:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_images(X):\n",
    "    plt.figure(1)\n",
    "    z = 0\n",
    "    for i in range(0,4):\n",
    "        for j in range(0,4):\n",
    "            plt.subplot2grid((4,4),(i,j))\n",
    "            plt.imshow(toimage(X[z]))\n",
    "            z = z + 1\n",
    "\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let us plot a few images:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 16 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_images(x_train[:17])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Defining Discriminator\n",
    "\n",
    "We define a discriminator $D$ as a convolutional network with three convolutional layers followed by a fully connected layer. It is composed of a series of convolutional and batch norm layers with Leaky ReLU activations. We apply batch normalization on all layers except at the input layer. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#initialize weights by drawing from a truncated normal distribution\n",
    "kernel_init = tf.truncated_normal_initializer(mean=0.0, stddev=0.02)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def discriminator(input_images, reuse=False, is_training=False, alpha=0.1):\n",
    "    \n",
    "    with tf.variable_scope('discriminator', reuse= reuse):\n",
    "        \n",
    "\n",
    "        #first convolutional layer with leaky relu activations\n",
    "        layer1 = tf.layers.conv2d(input_images, \n",
    "                                  filters=64, \n",
    "                                  kernel_size=5, \n",
    "                                  strides=2, \n",
    "                                  padding='same', \n",
    "                                  kernel_initializer=kernel_init, \n",
    "                                  name='conv1')\n",
    "\n",
    "        layer1 = tf.nn.leaky_relu(layer1, alpha=0.2, name='leaky_relu1')\n",
    "    \n",
    "        #second convolutional layer with batch normalization and leaky relu activations\n",
    "        layer2 = tf.layers.conv2d(layer1, \n",
    "                                  filters=128, \n",
    "                                  kernel_size=5, \n",
    "                                  strides=2, \n",
    "                                  padding='same', \n",
    "                                  kernel_initializer=kernel_init, \n",
    "                                  name='conv2')\n",
    "        layer2 = tf.layers.batch_normalization(layer2, training=is_training, name='batch_normalization2')\n",
    "        \n",
    "        layer2 = tf.nn.leaky_relu(layer2, alpha=0.2, name='leaky_relu2')\n",
    "\n",
    "        \n",
    "        #third convolutional layer with batch normalization followed by leaky relu activations\n",
    "        layer3 = tf.layers.conv2d(layer2, \n",
    "                                 filters=256, \n",
    "                                 kernel_size=5, \n",
    "                                 strides=1,\n",
    "                                 padding='same',\n",
    "                                 name='conv3')\n",
    "        layer3 = tf.layers.batch_normalization(layer3, training=is_training, name='batch_normalization3')\n",
    "        layer3 = tf.nn.leaky_relu(layer3, alpha=0.1, name='leaky_relu3')\n",
    "        \n",
    "        \n",
    "        #flatten the output of the final convolutional layer\n",
    "        layer3 = tf.reshape(layer3, (-1, layer3.shape[1]*layer3.shape[2]*layer3.shape[3]))\n",
    "        \n",
    "        \n",
    "        #define the fully connected layer and return the logits\n",
    "        logits= tf.layers.dense(layer3, 1)\n",
    "        \n",
    "        output = tf.sigmoid(logits)\n",
    "        \n",
    "        return logits"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Defining Generator\n",
    "\n",
    "\n",
    "As we learned generator $G$ in DCGAN performs the transpose convolutional operation.  The generator is composed of convolutional transpose and batch norm layers with ReLU activations.  We apply batch normalization on every layer expect at the last layer. Also, we apply ReLU activations on every layer but for the last layer, we apply tanh activation function to scale the generated image between -1 and +1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def generator(z, z_dim, batch_size, is_training=False, reuse=False):\n",
    "    \n",
    "    with tf.variable_scope('generator', reuse=reuse):\n",
    "        \n",
    "        #input\n",
    "        input_to_conv = tf.layers.dense(z, 8*8*128)\n",
    "        \n",
    "        #convert the shape of the input and apply batch normalization followed by a relu activations\n",
    "        layer1 = tf.reshape(input_to_conv, (-1, 8, 8, 128))\n",
    "        layer1 = tf.layers.batch_normalization(layer1, training=is_training, name='batch_normalization1')\n",
    "        layer1 = tf.nn.relu(layer1, name='relu1')\n",
    "        \n",
    "        #define the second layer which is the transpose convolution with batch normalization and relu activation\n",
    "        layer2 = tf.layers.conv2d_transpose(layer1, filters=256, kernel_size=5, strides= 2, padding='same', \n",
    "                                            kernel_initializer=kernel_init, name='deconvolution2')\n",
    "        layer2 = tf.layers.batch_normalization(layer2, training=is_training, name='batch_normalization2')\n",
    "        layer2 = tf.nn.relu(layer2, name='relu2')\n",
    "        \n",
    "        #define the third layer with transpose convolution operation with a batch norm and relu activation\n",
    "        layer3 = tf.layers.conv2d_transpose(layer2, filters=256, kernel_size=5, strides= 2, padding='same', \n",
    "                                            kernel_initializer=kernel_init, name='deconvolution3')\n",
    "        layer3 = tf.layers.batch_normalization(layer3,training=is_training, name='batch_normalization3')\n",
    "        layer3 = tf.nn.relu(layer3, name='relu3')\n",
    "        \n",
    "        #fourth layer  with convolutional transpose operation with a batch norm and relu activation\n",
    "        layer4 = tf.layers.conv2d_transpose(layer3, filters=256, kernel_size=5, strides= 1, padding='same', \n",
    "                                            kernel_initializer=kernel_init, name='deconvolution4')\n",
    "        layer4 = tf.layers.batch_normalization(layer4,training=is_training, name='batch_normalization4')\n",
    "        layer4 = tf.nn.relu(layer4, name='relu4')\n",
    "        \n",
    "        \n",
    "        #in the final layer, we don't apply batch normalization and instead of relu we use tanh activation\n",
    "        layer5 = tf.layers.conv2d_transpose(layer4, filters=3, kernel_size=7, strides=1, padding='same', \n",
    "                                            kernel_initializer=kernel_init, name='deconvolution5')\n",
    "           \n",
    "        \n",
    "        logits = tf.tanh(layer5, name='tanh')\n",
    "        \n",
    "        return logits"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Defining inputs\n",
    "\n",
    "Define the placeholder for our input $x$:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "image_width = x_train.shape[1]\n",
    "image_hight = x_train.shape[2]\n",
    "image_channels = x_train.shape[3]\n",
    "\n",
    "\n",
    "x = tf.placeholder(tf.float32, shape= (None, image_width, image_hight, image_channels), name=\"d_input\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define the placeholder for learning rate and training boolean:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learning_rate = tf.placeholder(tf.float32, shape=(), name=\"learning_rate\")\n",
    "is_training = tf.placeholder(tf.bool, [], name='is_training')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define the batch size and dimension of the noise $z$:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 100\n",
    "z_dim = 100"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define the placeholder for the noise z:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "z = tf.random_normal([batch_size, z_dim], mean=0.0, stddev=1.0, name='z')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Begin the GAN!\n",
    "\n",
    "\n",
    "First we feed the noise $z$ to the generator and it will output the fake image. i.e $ fake \\; x = G(z) $\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "fake_x = generator(z, z_dim, batch_size, is_training=is_training)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, we feed the real image to the discriminator $D(x)$ and get the probabillty of being real:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "D_logit_real = discriminator(x, reuse=False, is_training=is_training)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Similarly, we feed the fake image to the discirminator $D(z)$ and get the probabillty of it being real:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "D_logit_fake = discriminator(fake_x, reuse=True,  is_training=is_training)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Computing Loss Function\n",
    "\n",
    "Now, we will see, how to compute the loss function. It is just as SAME as we learned in GANs.\n",
    "\n",
    "\n",
    "\n",
    "### Discriminator Loss\n",
    "\n",
    "Discriminator loss is given as:\n",
    "\n",
    "${L ^{D} =  - \\mathbb{E}_{x \\sim p_{r}(x)}[\\log D(x)] - \\mathbb{E}_{z \\sim p_{z}(z)}[\\log (1-D(G(z))]} $\n",
    "\n",
    "\n",
    "<br>\n",
    "\n",
    "\n",
    "First, we will implement the first term i.e $\\mathbb{E}_{x \\sim p_{r}(x)}[\\log D(x)]$ "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "D_loss_real = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(labels=tf.ones_like(D_logit_real),\n",
    "                                                                     logits=D_logit_real))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, the second term, $\\mathbb{E}_{z \\sim p_{z}(z)}[\\log (1-D(G(z))]$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "D_loss_fake = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(labels=tf.zeros_like(D_logit_fake),\n",
    "                                                                     logits=D_logit_fake))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__Final term__\n",
    "\n",
    "So, combining the above two terms, loss function of the discriminator is given as:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "D_loss = D_loss_real + D_loss_fake"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Generator Loss\n",
    "\n",
    "Generator loss is given as:\n",
    "\n",
    "${L^{G}= - \\mathbb{E}_{z \\sim p_{z}(z)}[\\log (D(G(z)))] } $"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "G_loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(labels=tf.ones_like(D_logit_fake),\n",
    "                                                                logits=D_logit_fake))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Optimizing the Loss\n",
    "\n",
    "\n",
    "Now we need to optimize our generator and discriminator. So, we collect the parameters of the discriminator and generator as $\\theta_D$ and $\\theta_G$ respectively. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "training_vars = tf.trainable_variables()\n",
    "\n",
    "theta_D = [var for var in training_vars if var.name.startswith('discriminator')]\n",
    "theta_G = [var for var in training_vars if var.name.startswith('generator')]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Optimize the loss using adam optimizer:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "d_optimizer = tf.train.AdamOptimizer(learning_rate).minimize(D_loss, var_list=theta_D)\n",
    "g_optimizer = tf.train.AdamOptimizer(learning_rate).minimize(G_loss, var_list=theta_G)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Start the Training the DCGAN\n",
    "\n",
    "\n",
    "Define the number of batches, epochs and learning rate:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_batches = int(x_train.shape[0] / batch_size)\n",
    "steps = 0\n",
    "num_epcohs = 500\n",
    "lr = 0.00002"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define a helper function for generating and plotting images:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "def generate_new_samples(session, n_images, z_dim):\n",
    "    \n",
    "    #randomly sample noise from the normal distribution\n",
    "    z = tf.random_normal([1, z_dim], mean=0.0, stddev=1.0)\n",
    "\n",
    "    is_training = tf.placeholder(tf.bool, [], name='training_bool')     \n",
    "    \n",
    "    #generate a fake image\n",
    "    samples = session.run(generator(z, z_dim, batch_size, is_training, reuse=True),feed_dict={is_training: True})\n",
    "    \n",
    "    img = (samples[0] * 255).astype(np.uint8)\n",
    "    \n",
    "    #return the fake image\n",
    "    plt.imshow(img)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Generate new images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0, iteration: 0, Discriminator Loss:1.44706475735, Generator Loss: 0.726667642593\n"
     ]
    }
   ],
   "source": [
    "with tf.Session() as session:\n",
    "    \n",
    "    #initialize all variables\n",
    "    session.run(tf.global_variables_initializer())\n",
    "    \n",
    "    #for each epcohs\n",
    "    for epoch in range(num_epcohs):\n",
    "        \n",
    "        #for number of batches\n",
    "        for i in range(num_batches):\n",
    "            \n",
    "            #select start and end of the batch\n",
    "            start = i * batch_size\n",
    "            end = (i + 1) * batch_size\n",
    "            \n",
    "            #sample batch images\n",
    "            batch_images = x_train[start:end]\n",
    "            \n",
    "            #train the discriminator after every two steps\n",
    "            if(steps % 2 == 0):\n",
    "                \n",
    "                #train the discriminator\n",
    "                _, discriminator_loss = session.run([d_optimizer,D_loss], feed_dict={x: batch_images, is_training:True, learning_rate:lr})\n",
    "               \n",
    "            \n",
    "            #train the generator and discriminator\n",
    "            _, generator_loss = session.run([g_optimizer,G_loss], feed_dict={x: batch_images, is_training:True, learning_rate:lr})\n",
    "            _, discriminator_loss = session.run([d_optimizer,D_loss], feed_dict={x: batch_images, is_training:True, learning_rate:lr})\n",
    "            display.clear_output(wait=True)  \n",
    "            \n",
    "            #generate new image\n",
    "            generate_new_samples(session, 1, z_dim)\n",
    "            print(\"Epoch: {}, iteration: {}, Discriminator Loss:{}, Generator Loss: {}\".format(epoch,i,discriminator_loss,generator_loss))\n",
    "            \n",
    "\n",
    "            steps += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train the DCGAN for several epochs, so that you can see better images.\n",
    "\n",
    "\n",
    "Thus, we learned that DCGAN is similar to GANs except that it uses a convolutional network instead of a feedforward network in the generator and discriminator. In the next section, we will learn about LSGAN."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}