model.py

import tensorflow as tf
import ops
import subprocess as sb
import utils
from reader import Reader
from discriminator import Discriminator
from generator import Generator
import numpy as np

import vgg16

REAL_LABEL = 0.9

class CycleGAN:
  def __init__(self,
               X_train_file='',
               Y_train_file='',
               batch_size=1,
               image_size1=256,
               image_size2=256,
               use_lsgan=True,
               norm='instance',
               lambda1=10.0,
               lambda2=10.0,
               learning_rate=1e-4,
               beta1=0.5,
               ngf=64
              ):
    """
    Args:
      X_train_file: string, X tfrecords file for training
      Y_train_file: string Y tfrecords file for training
      batch_size: integer, batch size
      image_size: integer, image size
      lambda1: integer, weight for forward cycle loss (X->Y->X)
      lambda2: integer, weight for backward cycle loss (Y->X->Y)
      use_lsgan: boolean
      norm: 'instance' or 'batch'
      learning_rate: float, initial learning rate for Adam
      beta1: float, momentum term of Adam
      ngf: number of gen filters in first conv layer
    """
    self.lambda1 = lambda1
    self.lambda2 = lambda2
    self.use_lsgan = use_lsgan
    use_sigmoid = not use_lsgan
    self.batch_size = batch_size
    self.image_size1 = image_size1
    self.image_size2 = image_size2
    self.learning_rate = learning_rate
    self.beta1 = beta1
    self.X_train_file = X_train_file
    self.Y_train_file = Y_train_file

    self.is_training = tf.placeholder_with_default(True, shape=[], name='is_training')


    self.G = Generator('G', self.is_training, ngf=ngf, norm=norm, image_size1=image_size1, image_size2=image_size2)
    self.D_Y = Discriminator('D_Y',
        self.is_training, norm=norm, use_sigmoid=use_sigmoid)
    self.F = Generator('F', self.is_training, ngf=ngf, norm=norm, image_size1=image_size1, image_size2=image_size2)
    self.D_X = Discriminator('D_X',
        self.is_training, norm=norm, use_sigmoid=use_sigmoid)

    self.fake_x = tf.placeholder(tf.float32,
        shape=[batch_size, image_size1, image_size2, 3])
    self.fake_y = tf.placeholder(tf.float32,
        shape=[batch_size, image_size1, image_size2, 3])

    self.vgg = vgg16.Vgg16()

  def model(self):
    X_reader = Reader(self.X_train_file, name='X',
        image_size1=self.image_size1, image_size2=self.image_size2, batch_size=self.batch_size)
    Y_reader = Reader(self.Y_train_file, name='Y',
        image_size1=self.image_size1, image_size2=self.image_size2, batch_size=self.batch_size)

    x = X_reader.feed()
    y = Y_reader.feed()


    cycle_loss = self.cycle_consistency_loss(self.G, self.F, x, y)
    perceptual_loss = self.perceptual_similarity_loss(self.G, self.F, x, y, self.vgg)

    # X -> Y
    fake_y = self.G(x)
    G_gan_loss = self.generator_loss(self.D_Y, fake_y, use_lsgan=self.use_lsgan)
    G_loss = G_gan_loss + cycle_loss + perceptual_loss #+ pixel_loss
    D_Y_loss = self.discriminator_loss(self.D_Y, y, self.fake_y, use_lsgan=self.use_lsgan)

    # Y -> X
    fake_x = self.F(y)
    F_gan_loss = self.generator_loss(self.D_X, fake_x, use_lsgan=self.use_lsgan)
    F_loss = F_gan_loss + cycle_loss + perceptual_loss #+ pixel_loss
    D_X_loss = self.discriminator_loss(self.D_X, x, self.fake_x, use_lsgan=self.use_lsgan)


    # summary
    tf.summary.histogram('D_Y/true', self.D_Y(y))
    tf.summary.histogram('D_Y/fake', self.D_Y(self.G(x)))
    tf.summary.histogram('D_X/true', self.D_X(x))
    tf.summary.histogram('D_X/fake', self.D_X(self.F(y)))

    tf.summary.scalar('loss/G', G_gan_loss)
    tf.summary.scalar('loss/D_Y', D_Y_loss)
    tf.summary.scalar('loss/F', F_gan_loss)
    tf.summary.scalar('loss/D_X', D_X_loss)
    tf.summary.scalar('loss/cycle', cycle_loss)
    tf.summary.scalar('loss/perceptual_loss', perceptual_loss)
    #tf.summary.scalar('loss/pixel_loss', pixel_loss)

    tf.summary.image('X/generated', utils.batch_convert2int(self.G(x)))
    tf.summary.image('X/reconstruction', utils.batch_convert2int(self.F(self.G(x))))
    tf.summary.image('Y/generated', utils.batch_convert2int(self.F(y)))
    tf.summary.image('Y/reconstruction', utils.batch_convert2int(self.G(self.F(y))))

    return G_loss, D_Y_loss, F_loss, D_X_loss, fake_y, fake_x

  def optimize(self, G_loss, D_Y_loss, F_loss, D_X_loss):
    def make_optimizer(loss, variables, name='Adam'):
      """ Adam optimizer with learning rate 0.0002 for the first 100k steps (~100 epochs)
          and a linearly decaying rate that goes to zero over the next 100k steps
      """
      global_step = tf.Variable(0, trainable=False)
      starter_learning_rate = self.learning_rate
      end_learning_rate = 0.0
      start_decay_step = 100000
      decay_steps = 100000
      beta1 = self.beta1
      learning_rate = (
          tf.where(
                  tf.greater_equal(global_step, start_decay_step),
                  tf.train.polynomial_decay(starter_learning_rate, global_step-start_decay_step,
                                            decay_steps, end_learning_rate,
                                            power=1.0),
                  starter_learning_rate
          )

      )
      tf.summary.scalar('learning_rate/{}'.format(name), learning_rate)

      learning_step = (
          tf.train.AdamOptimizer(learning_rate, beta1=beta1, name=name)
                  .minimize(loss, global_step=global_step, var_list=variables)
      )
      return learning_step

    G_optimizer = make_optimizer(G_loss, self.G.variables, name='Adam_G')
    D_Y_optimizer = make_optimizer(D_Y_loss, self.D_Y.variables, name='Adam_D_Y')
    F_optimizer =  make_optimizer(F_loss, self.F.variables, name='Adam_F')
    D_X_optimizer = make_optimizer(D_X_loss, self.D_X.variables, name='Adam_D_X')
    with tf.control_dependencies([G_optimizer, D_Y_optimizer, F_optimizer, D_X_optimizer]):
      return tf.no_op(name='optimizers')

  def discriminator_loss(self, D, y, fake_y, use_lsgan=True):
    """ Note: default: D(y).shape == (batch_size,5,5,1),
                       fake_buffer_size=50, batch_size=1
    Args:
      G: generator object
      D: discriminator object
      y: 4D tensor (batch_size, image_size, image_size, 3)
    Returns:
      loss: scalar
    """
    if use_lsgan:
      # use mean squared error
      error_real = tf.reduce_mean(tf.squared_difference(D(y), REAL_LABEL))
      error_fake = tf.reduce_mean(tf.square(D(fake_y)))
    else:
      # use cross entropy
      error_real = -tf.reduce_mean(ops.safe_log(D(y)))
      error_fake = -tf.reduce_mean(ops.safe_log(1-D(fake_y)))
    loss = (error_real + error_fake) / 2
    return loss

  def generator_loss(self, D, fake_y, use_lsgan=True):
    """  fool discriminator into believing that G(x) is real
    """
    if use_lsgan:
      # use mean squared error
      loss = tf.reduce_mean(tf.squared_difference(D(fake_y), REAL_LABEL))
    else:
      # heuristic, non-saturating loss
      loss = -tf.reduce_mean(ops.safe_log(D(fake_y))) / 2
    return loss

  def cycle_consistency_loss(self, G, F, x, y):
    """ cycle consistency loss (L1 norm)
    """
    forward_loss = tf.reduce_mean(tf.abs(F(G(x))-x))
    backward_loss = tf.reduce_mean(tf.abs(G(F(y))-y))
    loss = self.lambda1*forward_loss + self.lambda2*backward_loss
    return loss

  def perceptual_similarity_loss(self, G, F, x, y, vgg):
    x1 = tf.image.resize_images(x, [224,224]) # to feed vgg, need resize
    y1 = tf.image.resize_images(y, [224,224])

    rx = F(G(x)) #create reconstructed images
    ry = G(F(y))

    rx1 = tf.image.resize_images(rx, [224,224]) # to feed vgg, need resize
    ry1 = tf.image.resize_images(ry, [224,224])

    fx1, fx2 = vgg.build(x1) # extract features from vgg
    fy1, fy2 = vgg.build(y1)

    frx1, frx2 = vgg.build(rx1) # extract features from vgg (2nd pool & 5th pool
    fry1, fry2 = vgg.build(ry1)

    m1 = tf.reduce_mean(tf.squared_difference(fx1, frx1)) # mse difference
    m2 = tf.reduce_mean(tf.squared_difference(fx2, frx2))

    m3 = tf.reduce_mean(tf.squared_difference(fy1, fry1))
    m4 = tf.reduce_mean(tf.squared_difference(fy2, fry2))

    perceptual_loss = (m1 + m2 + m3 + m4)*0.00001*0.5 # calculate perceptual loss and give weight (0.00001*0.5)
    return perceptual_loss

 # def pixel_wise_loss(self, G, F, x, y):
 #   rx = F(G(x))
 #   ry = G(F(y))
 #   pixel_wise_loss = tf.reduce_mean(tf.squared_difference(x, rx)) + tf.reduce_mean(tf.squared_difference(y, ry))
 #   return 10*pixel_wise_loss