def make_generator_model():
model = tf.keras.Sequential()
model.add(Dense(7 * 7 * 256, use_bias=False, input_shape=(100, )))
model.add(BatchNormalization())
model.add(Activation("relu"))
model.add(Reshape((7, 7, 256)))
model.add(C2DT(128, (5, 5), strides=(1, 1), padding='same',
use_bias=False))
model.add(BatchNormalization())
model.add(Activation("relu"))
model.add(C2DT(64, (5, 5), strides=(2, 2), padding='same', use_bias=False))
model.add(BatchNormalization())
model.add(Activation("relu"))
model.add(
C2DT(3, (5, 5),
strides=(2, 2),
padding='same',
use_bias=False,
activation='tanh'))
assert model.output_shape == (None, 28, 28, 3)
print("========\ngenerator")
print(model.summary())
return model
def make_generator_model():
model = tf.keras.Sequential()
init = RandomNormal(mean=0.0, stddev=0.02)
model.add(Dense(7 * 7 * 128, kernel_initializer=init, input_shape=(100, )))
model.add(Reshape((7, 7, 128)))
model.add(BatchNormalization())
model.add(LeakyReLU(alpha=0.2))
model.add(C2DT(64, 5, strides=2, padding='same', kernel_initializer=init))
model.add(BatchNormalization())
model.add(LeakyReLU(alpha=0.2))
# model.add(C2DT(32, 5, strides=2, padding='same', kernel_initializer=init))
# model.add(BatchNormalization())
# model.add(LeakyReLU(alpha=0.2))
model.add(
C2DT(3,
5,
strides=2,
padding='same',
activation='tanh',
kernel_initializer=init))
print(model.summary())
assert model.output_shape == (None, 28, 28, 3)
return model
def make_generator_model():
model = tf.keras.Sequential()
# start with 4x4 image
model.add(Dense(4 * 4 * 256, use_bias=False, input_dim=100))
#model.add(BatchNormalization())
model.add(LeakyReLU())
model.add(Reshape((4, 4, 256)))
# upsample to 8x8
model.add(C2DT(128, (4, 4), strides=(2, 2), padding='same',
use_bias=False))
#model.add(BatchNormalization())
model.add(LeakyReLU())
# upsample to 16x16
model.add(C2DT(128, (4, 4), strides=(2, 2), padding='same',
use_bias=False))
#model.add(BatchNormalization())
model.add(LeakyReLU())
# upsample to 32x32
model.add(C2DT(128, (4, 4), strides=(2, 2), padding='same',
use_bias=False))
#model.add(BatchNormalization())
model.add(LeakyReLU())
# output layer
model.add(Conv2D(3, (3, 3), activation='tanh', padding='same'))
assert model.output_shape == (None, 32, 32, 3)
return model
def make_generator_model():
model = tf.keras.Sequential()
model.add(Dense(7*7*256, use_bias=False, input_shape=(100,)))
model.add(BatchNormalization())
model.add(LeakyReLU())
model.add(Reshape((7, 7, 256)))
model.add(C2DT(128, (5, 5), strides=(1, 1), padding='same', use_bias=False))
model.add(BatchNormalization())
model.add(LeakyReLU())
model.add(C2DT(64, (5, 5), strides=(2, 2), padding='same', use_bias=False))
model.add(BatchNormalization())
model.add(LeakyReLU())
model.add(C2DT(1, (5, 5), strides=(2, 2), padding='same', use_bias=False, activation='tanh'))
assert model.output_shape == (None, 56, 56, 1)
return model
def make_generator_model():
# changed activation func relu to leakyrelu. https://github.com/soumith/ganhacks#6-use-soft-and-noisy-labels tip #5
# add gaussian weight init
model = tf.keras.Sequential()
init = RandomNormal(mean=0.0, stddev=0.02)
model.add(Dense(7 * 7 * 128, kernel_initializer=init, input_shape=(100, )))
model.add(Reshape((7, 7, 128))) # prøv med 256
model.add(BatchNormalization())
model.add(LeakyReLU(alpha=0.2))
model.add(
C2DT(64, (5, 5),
strides=(2, 2),
padding='same',
kernel_initializer=init))
model.add(BatchNormalization())
model.add(LeakyReLU(alpha=0.2))
model.add(
C2DT(32, (5, 5),
strides=(2, 2),
padding='same',
kernel_initializer=init))
model.add(BatchNormalization())
model.add(LeakyReLU(alpha=0.2))
model.add(
C2DT(1, (5, 5),
strides=(2, 2),
padding='same',
activation='tanh',
kernel_initializer=init))
print(model.summary())
assert model.output_shape == (None, 56, 56, 1)
return model
def define_generator_model(k):
model = Sequential() # 初始化一个序列（网络），接下里会向序列中add层
#全连接层，
model.add(Dense(7 * 7 * 256, use_bias=False, input_shape=(k, )))
model.add(BatchNormalization())
model.add(LeakyReLU()) # 添加了一个很小的γ斜率,避免x<0的永远不会被激活
#
model.add(Reshape((7, 7, 256)))
assert model.output_shape == (None, 7, 7, 256) # 不限制batch size
#第一个转置卷积层，s=1
model.add(C2DT(128, (5, 5), strides=(1, 1), padding='same',
use_bias=False))
assert model.output_shape == (None, 7, 7, 128)
model.add(BatchNormalization())
model.add(LeakyReLU())
#第二个转置卷积层，s=2,做0填充
model.add(C2DT(64, (5, 5), strides=(2, 2), padding='same', use_bias=False))
assert model.output_shape == (None, 14, 14, 64)
model.add(BatchNormalization())
model.add(LeakyReLU())
#第三个转置卷积层，filter=32,kernal_size=5,s=2,做0填充
model.add(C2DT(32, (5, 5), strides=(2, 2), padding='same', use_bias=False))
assert model.output_shape == (None, 28, 28, 32)
model.add(BatchNormalization())
model.add(LeakyReLU())
#第四个转置卷积层，
model.add(C2DT(16, (5, 5), strides=(2, 2), padding='same', use_bias=False))
assert model.output_shape == (None, 56, 56, 16)
model.add(BatchNormalization())
model.add(LeakyReLU())
#第五个转置卷积层，filter=1,kernal_size=5,s=2,做0填充
model.add(
C2DT(1, (5, 5),
strides=(2, 2),
padding='same',
use_bias=False,
activation='tanh'))
assert model.output_shape == (None, 112, 112, 1)
return model
def make_generator_model():
model = Sequential()
init = RandomNormal(mean=0.0, stddev=0.02)
model.add(
Dense(7 * 7 * 128,
kernel_initializer=init,
use_bias=False,
input_shape=(100, )))
model.add(Reshape((7, 7, 128)))
model.add(BatchNormalization())
model.add(LeakyReLU(alpha=0.2))
model.add(
C2DT(64, (5, 5),
strides=(2, 2),
kernel_initializer=init,
padding='same'))
model.add(BatchNormalization())
model.add(LeakyReLU(alpha=0.2))
# model.add(C2DT(64, (5, 5), strides=(2, 2), kernel_initializer=init, padding='same'))
# model.add(BatchNormalization())
# model.add(LeakyReLU(alpha=0.2))
# model.add(C2DT(32, (5, 5), strides=(2, 2), kernel_initializer=init, padding='same'))
# model.add(BatchNormalization())
# model.add(LeakyReLU(alpha=0.2))
model.add(
C2DT(1, (5, 5),
strides=(2, 2),
padding='same',
kernel_initializer=init,
activation='tanh'))
print(model.output_shape)
assert model.output_shape == (None, 28, 28, 1)
return model