def define_model(self, images_x, images_y):
"""Defines a CycleGAN model that maps between images_x and images_y.
Args:
images_x: A 4D float `Tensor` of NHWC format. Images in set X.
images_y: A 4D float `Tensor` of NHWC format. Images in set Y.
use_identity_loss: Whether to use identity loss or not
Returns:
A `CycleGANModel` namedtuple.
"""
if self._use_identity_loss:
cyclegan_model = cyclegan_model_with_identity(
generator_fn=_shadowdata_generator_model,
discriminator_fn=_shadowdata_discriminator_model,
data_x=images_x,
data_y=images_y)
else:
cyclegan_model = tfgan.cyclegan_model(
generator_fn=_shadowdata_generator_model,
discriminator_fn=_shadowdata_discriminator_model,
data_x=images_x,
data_y=images_y)
# Add summaries for generated images.
# tfgan.eval.add_cyclegan_image_summaries(cyclegan_model)
return cyclegan_model
def _define_model(images_x, images_y):
"""Defines a CycleGAN model that maps between images_x and images_y.
Args:
images_x: A 4D float `Tensor` of NHWC format. Images in set X.
images_y: A 4D float `Tensor` of NHWC format. Images in set Y.
Returns:
A `CycleGANModel` namedtuple.
"""
cyclegan_model = tfgan.cyclegan_model(
generator_fn=networks.generator,
discriminator_fn=networks.discriminator,
data_x=images_x,
data_y=images_y)
# Add summaries for generated images.
tfgan.eval.add_cyclegan_image_summaries(cyclegan_model)
return cyclegan_model
def create_callable_cyclegan_model():
return tfgan.cyclegan_model(
Generator(),
Discriminator(),
data_x=tf.zeros([1, 2]),
data_y=tf.ones([1, 2]))
def cyclegan_model_with_identity(
# Lambdas defining models.
generator_fn,
discriminator_fn,
# data X and Y.
data_x,
data_y,
# Optional scopes.
generator_scope="Generator",
discriminator_scope="Discriminator",
model_x2y_scope="ModelX2Y",
model_y2x_scope="ModelY2X",
# Options.
check_shapes=True):
"""Returns a CycleGAN model outputs and variables.
See https://arxiv.org/abs/1703.10593 for more details.
Args:
generator_fn: A python lambda that takes `data_x` or `data_y` as inputs and
returns the outputs of the GAN generator.
discriminator_fn: A python lambda that takes `real_data`/`generated data`
and `generator_inputs`. Outputs a Tensor in the range [-inf, inf].
data_x: A `Tensor` of dataset X. Must be the same shape as `data_y`.
data_y: A `Tensor` of dataset Y. Must be the same shape as `data_x`.
generator_scope: Optional generator variable scope. Useful if you want to
reuse a subgraph that has already been created. Defaults to 'Generator'.
discriminator_scope: Optional discriminator variable scope. Useful if you
want to reuse a subgraph that has already been created. Defaults to
'Discriminator'.
model_x2y_scope: Optional variable scope for model x2y variables. Defaults
to 'ModelX2Y'.
model_y2x_scope: Optional variable scope for model y2x variables. Defaults
to 'ModelY2X'.
check_shapes: If `True`, check that generator produces Tensors that are the
same shape as `data_x` (`data_y`). Otherwise, skip this check.
Returns:
A `CycleGANModel` namedtuple.
Raises:
ValueError: If `check_shapes` is True and `data_x` or the generator output
does not have the same shape as `data_y`.
ValueError: If TF is executing eagerly.
"""
original_model = tfgan.cyclegan_model(
generator_fn=generator_fn,
discriminator_fn=discriminator_fn,
data_x=data_x,
data_y=data_y,
generator_scope=generator_scope,
discriminator_scope=discriminator_scope,
model_x2y_scope=model_x2y_scope,
model_y2x_scope=model_y2x_scope,
check_shapes=check_shapes)
with tf.compat.v1.variable_scope(original_model.model_x2y.generator_scope,
reuse=True):
identity_x = original_model.model_x2y.generator_fn(data_x)
with tf.compat.v1.variable_scope(original_model.model_y2x.generator_scope,
reuse=True):
identity_y = original_model.model_y2x.generator_fn(data_y)
model_w_identity = CycleGANModelWithIdentity(
model_x2y=original_model.model_x2y,
model_y2x=original_model.model_y2x,
reconstructed_x=original_model.reconstructed_x,
reconstructed_y=original_model.reconstructed_y)
model_w_identity.identity_x = identity_x
model_w_identity.identity_y = identity_y
return model_w_identity
def create_cyclegan_model():
return tfgan.cyclegan_model(
generator_model,
discriminator_model,
data_x=tf.zeros([1, 2]),
data_y=tf.ones([1, 2]))