def test_generator_inference(self):
"""Check one inference step."""
img_batch = tf.zeros([2, 32, 32, 3])
model_output, _ = generator.cyclegan_generator_resnet(img_batch)
with self.cached_session() as sess:
sess.run(tf.compat.v1.global_variables_initializer())
sess.run(model_output)
def input_and_output_same_shape(self, kernel_size):
img_batch = tf.compat.v1.placeholder(tf.float32, shape=[None, 32, 32, 3])
output_img_batch, _ = generator.cyclegan_generator_resnet(
img_batch, kernel_size=kernel_size)
self.assertAllEqual(img_batch.shape.as_list(),
output_img_batch.shape.as_list())
def test_generator_unknown_batch_dim(self):
"""Check that generator can take unknown batch dimension inputs."""
if tf.executing_eagerly():
# tf.placeholder() is not compatible with eager execution.
return
img = tf.compat.v1.placeholder(tf.float32, shape=[None, 32, None, 3])
output_imgs, _ = generator.cyclegan_generator_resnet(img)
self.assertAllEqual([None, 32, None, 3], output_imgs.shape.as_list())
def generator(input_images):
"""Thin wrapper around CycleGAN generator to conform to the TF-GAN API.
Args:
input_images: A batch of images to translate. Images should be normalized
already. Shape is [batch, height, width, channels].
Returns:
Returns generated image batch.
Raises:
ValueError: If shape of last dimension (channels) is not defined.
"""
input_images.shape.assert_has_rank(4)
input_size = input_images.shape.as_list()
channels = input_size[-1]
if channels is None:
raise ValueError('Last dimension shape must be known but is None: %s' %
input_size)
output_images, _ = gmodule.cyclegan_generator_resnet(input_images)
return output_images
def generator(input_images):
"""Thin wrapper around CycleGAN generator to conform to the TF-GAN API.
Args:
input_images: A batch of images to translate. Images should be normalized
already. Shape is [batch, height, width, channels].
Returns:
Returns generated image batch.
Raises:
ValueError: If shape of last dimension (channels) is not defined.
"""
input_images.shape.assert_has_rank(4)
input_size = input_images.shape.as_list()
channels = input_size[-1]
if channels is None:
raise ValueError('Last dimension shape must be known but is None: %s' %
input_size)
output_images, _ = gmodule.cyclegan_generator_resnet(input_images,
tanh_linear_slope=0.1)
# Optionally add image to summaries.
# tf.summary.image('generator_preconcat_residue', output_images)
# Difference between cycleGAN and the version used for DME:
# 1. We have a 1 × 1 convolutional path from the input to the output.
concat_images = tf.concat([output_images, input_images], axis=3)
output_images = concat_images
output_images = tf.layers.conv2d(output_images,
channels, [1, 1],
activation=None)
# Optionally add image to summaries.
# tf.summary.image('generator_residue', output_images)
# 2. We model this function as a residual.
output_images += input_images
return output_images
def test_generator_graph(self, shape):
"""Check that generator can take small and non-square inputs."""
output_imgs, _ = generator.cyclegan_generator_resnet(tf.ones(shape))
self.assertAllEqual(shape, output_imgs.shape.as_list())
def test_generator_unknown_batch_dim(self):
"""Check that generator can take unknown batch dimension inputs."""
img = tf.compat.v1.placeholder(tf.float32, shape=[None, 32, None, 3])
output_imgs, _ = generator.cyclegan_generator_resnet(img)
self.assertAllEqual([None, 32, None, 3], output_imgs.shape.as_list())
