def test_generator_run(self):
tf.set_random_seed(1234)
noise = tf.random_normal([100, 64])
image, _ = dcgan.generator(noise)
with self.test_session() as sess:
sess.run(tf.global_variables_initializer())
image.eval()
def test_generator_run(self):
tf.set_random_seed(1234)
noise = tf.random_normal([100, 64])
image, _ = dcgan.generator(noise)
with self.test_session() as sess:
sess.run(tf.global_variables_initializer())
image.eval()
def test_generator_invalid_input(self):
wrong_dim_input = tf.zeros([5, 32, 32])
with self.assertRaises(ValueError):
dcgan.generator(wrong_dim_input)
correct_input = tf.zeros([3, 2])
with self.assertRaisesRegexp(ValueError, 'must be a power of 2'):
dcgan.generator(correct_input, final_size=30)
with self.assertRaisesRegexp(ValueError, 'must be greater than 8'):
dcgan.generator(correct_input, final_size=4)
def test_generator_invalid_input(self):
wrong_dim_input = tf.zeros([5, 32, 32])
with self.assertRaises(ValueError):
dcgan.generator(wrong_dim_input)
correct_input = tf.zeros([3, 2])
with self.assertRaisesRegexp(ValueError, 'must be a power of 2'):
dcgan.generator(correct_input, final_size=30)
with self.assertRaisesRegexp(ValueError, 'must be greater than 8'):
dcgan.generator(correct_input, final_size=4)
def test_generator_graph(self):
tf.set_random_seed(1234)
# Check graph construction for a number of image size/depths and batch
# sizes.
for i, batch_size in zip(xrange(3, 7), xrange(3, 8)):
tf.reset_default_graph()
final_size = 2 ** i
noise = tf.random_normal([batch_size, 64])
image, end_points = dcgan.generator(
noise,
depth=32,
final_size=final_size)
self.assertAllEqual([batch_size, final_size, final_size, 3],
image.shape.as_list())
expected_names = ['deconv%i' % j for j in xrange(1, i)] + ['logits']
self.assertSetEqual(set(expected_names), set(end_points.keys()))
# Check layer depths.
for j in range(1, i):
layer = end_points['deconv%i' % j]
self.assertEqual(32 * 2**(i-j-1), layer.get_shape().as_list()[-1])
def test_generator_graph(self):
tf.set_random_seed(1234)
# Check graph construction for a number of image size/depths and batch
# sizes.
for i, batch_size in zip(xrange(3, 7), xrange(3, 8)):
tf.reset_default_graph()
final_size = 2 ** i
noise = tf.random_normal([batch_size, 64])
image, end_points = dcgan.generator(
noise,
depth=32,
final_size=final_size)
self.assertAllEqual([batch_size, final_size, final_size, 3],
image.shape.as_list())
expected_names = ['deconv%i' % j for j in xrange(1, i)] + ['logits']
self.assertSetEqual(set(expected_names), set(end_points.keys()))
# Check layer depths.
for j in range(1, i):
layer = end_points['deconv%i' % j]
self.assertEqual(32 * 2**(i-j-1), layer.get_shape().as_list()[-1])
(epoch_size + 1)))
D_model.save_weights('logs/D_Epoch%d-GLoss%.4f-DLoss%.4f.h5' %
((epoch + 1), G_total_loss /
(epoch_size + 1), D_total_loss /
(epoch_size + 1)))
print('Saving state, iter:', str(epoch + 1))
if __name__ == "__main__":
image_shape = [64, 64, 3]
# 数据集存放路径
annotation_path = "train_lines.txt"
# 生成网络和评价网络
G_model = generator(64, image_shape)
D_model = discriminator(64, image_shape)
# G_model_path = "model_data/Generator_Flower.h5"
# D_model_path = "model_data/Discriminator_Flower.h5"
# G_model.load_weights(G_model_path, by_name=True, skip_mismatch=True)
# D_model.load_weights(D_model_path, by_name=True, skip_mismatch=True)
with open(annotation_path) as f:
lines = f.readlines()
num_train = len(lines)
#------------------------------------------------------#
# Init_Epoch为起始世代
# Epoch总训练世代
#------------------------------------------------------#
Epoch = 500
batch_size = 64
lr = 0.002
#------------------------------#
# 每隔50个step保存一次图片
#------------------------------#
save_interval = 50
#------------------------------------------#
# 获得图片路径
#------------------------------------------#
annotation_path = "train_lines.txt"
#----------------------------#
# 生成网络和评价网络
#----------------------------#
G_model = generator(channel, input_shape)
D_model = discriminator(channel, input_shape)
#------------------------------------------#
# 将训练好的模型重新载入
#------------------------------------------#
if G_model_path != '':
G_model.load_weights(G_model_path, by_name=True, skip_mismatch=True)
if D_model_path != '':
D_model.load_weights(D_model_path, by_name=True, skip_mismatch=True)
with open(annotation_path) as f:
lines = f.readlines()
num_train = len(lines)
#------------------------------------------------------#
def generate(self):
self.net = generator(self.channel, self.image_shape)
self.net.load_weights(self.model_path)
print('{} model loaded.'.format(self.model_path))
