def discriminate(self, conv, reuse=False, pg=1, t=False, alpha_trans=0.01):
#dis_as_v = []
with tf.variable_scope("discriminator") as scope:
if reuse == True:
scope.reuse_variables()
if t:
conv_iden = avgpool2d(conv)
#from RGB
conv_iden = lrelu(conv2d(conv_iden, output_dim= self.get_nf(pg - 2), k_w=1, k_h=1, d_h=1, d_w=1,
name='dis_y_rgb_conv_{}'.format(conv_iden.shape[1])))
# fromRGB
conv = lrelu(conv2d(conv, output_dim=self.get_nf(pg - 1), k_w=1, k_h=1, d_w=1, d_h=1, name='dis_y_rgb_conv_{}'.format(conv.shape[1])))
for i in range(pg - 1):
conv = lrelu(conv2d(conv, output_dim=self.get_nf(pg - 1 - i), d_h=1, d_w=1,
name='dis_n_conv_1_{}'.format(conv.shape[1])))
conv = lrelu(conv2d(conv, output_dim=self.get_nf(pg - 2 - i), d_h=1, d_w=1,
name='dis_n_conv_2_{}'.format(conv.shape[1])))
conv = avgpool2d(conv, 2)
if i == 0 and t:
conv = alpha_trans * conv + (1 - alpha_trans) * conv_iden
conv = MinibatchstateConcat(conv)
conv = lrelu(
conv2d(conv, output_dim=self.get_nf(1), k_w=3, k_h=3, d_h=1, d_w=1, name='dis_n_conv_1_{}'.format(conv.shape[1])))
conv = lrelu(
conv2d(conv, output_dim=self.get_nf(1), k_w=4, k_h=4, d_h=1, d_w=1, padding='VALID', name='dis_n_conv_2_{}'.format(conv.shape[1])))
conv = tf.reshape(conv, [self.batch_size, -1])
#for D
output = fully_connect(conv, output_size=1, scope='dis_n_fully')
return tf.nn.sigmoid(output), output
def discriminate(self, conv, t_text_embedding, reuse=False, pg=1, t=False, alpha_trans=0.01):
#NOTE: discriminate from PGGAN does not use batch norm, add later?
#dis_as_v = []
with tf.variable_scope("discriminator") as scope:
if reuse == True:
scope.reuse_variables()
if t:
conv_iden = avgpool2d(conv)
#from RGB
conv_iden = lrelu(conv2d(conv_iden, output_dim= self.get_nf(pg - 2), k_w=1, k_h=1, d_h=1, d_w=1,
name='dis_y_rgb_conv_{}'.format(conv_iden.shape[1])))
# fromRGB
conv = lrelu(conv2d(conv, output_dim=self.get_nf(pg - 1), k_w=1, k_h=1, d_w=1, d_h=1, name='dis_y_rgb_conv_{}'.format(conv.shape[1])))
for i in range(pg - 1):
conv = lrelu(conv2d(conv, output_dim=self.get_nf(pg - 1 - i), d_h=1, d_w=1,
name='dis_n_conv_1_{}'.format(conv.shape[1])))
conv = lrelu(conv2d(conv, output_dim=self.get_nf(pg - 2 - i), d_h=1, d_w=1,
name='dis_n_conv_2_{}'.format(conv.shape[1])))
conv = avgpool2d(conv, 2)
if i == 0 and t:
conv = alpha_trans * conv + (1 - alpha_trans) * conv_iden
conv = MinibatchstateConcat(conv)
conv = lrelu(
conv2d(conv, output_dim=self.get_nf(1), k_w=3, k_h=3, d_h=1, d_w=1, name='dis_n_conv_1_{}'.format(conv.shape[1])))
conv = lrelu(
conv2d(conv, output_dim=self.get_nf(1), k_w=4, k_h=4, d_h=1, d_w=1, padding='VALID', name='dis_n_conv_2_{}'.format(conv.shape[1])))
#ADD TEXT EMBEDDING TO THE NETWORK
reduced_text_embeddings = lrelu(linear(t_text_embedding, self.tdim, 'd_embedding'))
reduced_text_embeddings = tf.expand_dims(reduced_text_embeddings,1)
reduced_text_embeddings = tf.expand_dims(reduced_text_embeddings,2)
#NOTE: output of prev layer is a 1x1 volume, so we don't tile by 4
tiled_embeddings = tf.tile(reduced_text_embeddings, [1,1,1,1], name='tiled_embeddings') #last conv layer op should be 4x4
conv_concat = tf.concat( [conv, tiled_embeddings], 3, name='dis_conv_concat_{}'.format(conv.shape[1]))
#NOTE: changed the output dims here as compared to text to image code
conv_new = lrelu((conv2d(conv_concat, output_dim=self.get_nf(1), k_h=1,k_w=1,d_h=1,d_w=1, name = 'dis_conv_new_{}'.format(conv_concat.shape[1])))) #4
conv_new = tf.reshape(conv_new, [self.batch_size, -1])
#for D
output = fully_connect(conv_new, output_size=1, scope='dis_n_fully')
return tf.nn.sigmoid(output), output
def __call__(self,
x,
is_reuse=False,
is_train=True,
is_intermediate=False):
with tf.variable_scope('generator') as scope:
if is_reuse:
scope.reuse_variables()
unit_size = self.img_size[0] // (2**self.layer_n)
unit_n = self.smallest_hidden_unit_n * (2**(self.layer_n - 1))
batch_size = int(x.shape[0])
if is_intermediate:
intermediate_xs = list()
with tf.variable_scope('pre'):
x = linear(x, unit_size * unit_size * unit_n)
x = tf.reshape(x, (batch_size, unit_size, unit_size, unit_n))
if self.is_bn:
x = batch_norm(x, is_train)
x = tf.nn.relu(x)
if is_intermediate:
y = avgpool2d(x, unit_size, 1, 'VALID')
y = tf.reshape(y, (batch_size, -1))
intermediate_xs.append(y)
for i in range(self.layer_n):
with tf.variable_scope('layer{}'.format(i)):
if i == self.layer_n - 1:
unit_n = self.img_dim
else:
unit_n = self.smallest_hidden_unit_n * (2**(
self.layer_n - i - 2))
x_shape = x.get_shape().as_list()
if self.is_transpose:
x = deconv2d(x, [
x_shape[0], x_shape[1] * 2, x_shape[1] * 2, unit_n
], self.k_size, 2, 'SAME')
else:
x = tf.image.resize_bilinear(
x, (x_shape[1] * 2, x_shape[2] * 2))
x = conv2d(x, unit_n, self.k_size, 1, 'SAME')
if i != self.layer_n - 1:
if self.is_bn:
x = batch_norm(x, is_train)
x = tf.nn.relu(x)
if is_intermediate:
y = tf.reshape(x, (batch_size, -1))
intermediate_xs.append(y)
x = tf.nn.tanh(x)
if is_intermediate:
return x, intermediate_xs
return x
def discriminate(self, conv, reuse=False, pg=1, t=False, alpha_trans=0.01):
#dis_as_v = []
with tf.variable_scope("discriminator") as scope:
if reuse == True:
scope.reuse_variables()
# transition이 True라면(즉, 해상도를 두배로 할 시에)
if t:
conv_iden = avgpool2d(conv)
#from RGB
# pg=2라면 get_nf에서 512가 나옴
# name은 dis_y_rgb_conv_(현재 처리하는 해상도/2)
conv_iden = lrelu(
conv2d(conv_iden,
output_dim=self.get_nf(pg - 2),
k_w=1,
k_h=1,
d_h=1,
d_w=1,
name='dis_y_rgb_conv_{}'.format(
conv_iden.shape[1])))
# fromRGB
# pg=1일때에 get_nf(pg-1)에서 512 나옴
# pg=2일때에 get_nf(pg-1)에서 512 나옴
# pg=3일때에 get_nf(pg-1)에서 256 나옴
# name은 dis_y_rgb_conv_(현재 처리하는 해상도)
conv = lrelu(
conv2d(conv,
output_dim=self.get_nf(pg - 1),
k_w=1,
k_h=1,
d_w=1,
d_h=1,
name='dis_y_rgb_conv_{}'.format(conv.shape[1])))
# pg=1 일때는 for문 돌지 않음
# pg=2 일때 i=0나옴
# pg=3 일때 i=0, 1 나옴
for i in range(pg - 1):
# 기본 kernel size는 3x3
# pg=2이고 i=0일때 get_nf(pg-1-i)에서 512가 나옴
conv = lrelu(
conv2d(conv,
output_dim=self.get_nf(pg - 1 - i),
d_h=1,
d_w=1,
name='dis_n_conv_1_{}'.format(conv.shape[1])))
# pg=2 이고 i=0일때 get_nf(pg-2-i)에서 512가 나나옴
conv = lrelu(
conv2d(conv,
output_dim=self.get_nf(pg - 2 - i),
d_h=1,
d_w=1,
name='dis_n_conv_2_{}'.format(conv.shape[1])))
conv = avgpool2d(conv, 2)
# 첫번째 block이고 transition이 True라면 Blending시킴
if i == 0 and t:
conv = alpha_trans * conv + (1 - alpha_trans) * conv_iden
conv = MinibatchstateConcat(conv)
# 무조건 channel dimension 512
conv = lrelu(
conv2d(conv,
output_dim=self.get_nf(1),
k_w=3,
k_h=3,
d_h=1,
d_w=1,
name='dis_n_conv_1_{}'.format(conv.shape[1])))
# 무조건 channel dimension 512, VALID라서 feature map의 width, height가 줄어듬
conv = lrelu(
conv2d(conv,
output_dim=self.get_nf(1),
k_w=4,
k_h=4,
d_h=1,
d_w=1,
padding='VALID',
name='dis_n_conv_2_{}'.format(conv.shape[1])))
# 1차원으로 reshape
conv = tf.reshape(conv, [self.batch_size, -1])
#for D
output = fully_connect(conv, output_size=1, scope='dis_n_fully')
return tf.nn.sigmoid(output), output
def discriminate(self,
input_image,
reuse=False,
model_progressive_depth=1,
transition=False,
alpha_transition=0.01,
input_classes=None):
with tf.variable_scope("discriminator") as scope:
if reuse:
scope.reuse_variables()
if transition:
# from RGB, low resolution
transition_conv = avgpool2d(input_image)
transition_conv = lrelu(
conv2d(transition_conv,
output_dim=self.get_filter_num(
model_progressive_depth - 2),
k_w=1,
k_h=1,
d_h=1,
d_w=1,
name='dis_y_rgb_conv_{}'.format(
transition_conv.shape[1])))
convs = []
# from RGB, high resolution
convs += [
lrelu(
conv2d(input_image,
output_dim=self.get_filter_num(
model_progressive_depth - 1),
k_w=1,
k_h=1,
d_w=1,
d_h=1,
name='dis_y_rgb_conv_{}'.format(
input_image.shape[1])))
]
for i in range(model_progressive_depth - 1):
convs += [
lrelu(
conv2d(convs[-1],
output_dim=self.get_filter_num(
model_progressive_depth - 1 - i),
d_h=1,
d_w=1,
name='dis_n_conv_1_{}'.format(
convs[-1].shape[1])))
]
convs += [
lrelu(
conv2d(convs[-1],
output_dim=self.get_filter_num(
model_progressive_depth - 2 - i),
d_h=1,
d_w=1,
name='dis_n_conv_2_{}'.format(
convs[-1].shape[1])))
]
convs[-1] = avgpool2d(convs[-1], 2)
if i == 0 and transition:
convs[-1] = alpha_transition * convs[-1] + (
1 - alpha_transition) * transition_conv
convs += [minibatch_state_concat(convs[-1])]
convs[-1] = lrelu(
conv2d(convs[-1],
output_dim=self.get_filter_num(1),
k_w=3,
k_h=3,
d_h=1,
d_w=1,
name='dis_n_conv_1_{}'.format(convs[-1].shape[1])))
output = tf.reshape(convs[-1], [self.batch_size, -1])
discriminate_output = fully_connect(output,
output_size=1,
scope='dis_n_fully')
return tf.nn.sigmoid(discriminate_output), discriminate_output