-
Notifications
You must be signed in to change notification settings - Fork 0
/
model.py
234 lines (191 loc) · 9.59 KB
/
model.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
'''
code based on https://github.com/xhujoy/CycleGAN-tensorflow/blob/master/model.py
'''
from __future__ import division
import os
import numpy as np
from keras.layers import Input
from keras.models import Model
from keras.optimizers import Adam
import datetime
import matplotlib.pyplot as plt
from data_loader import DataLoader
from module import *
from utils import *
from option import *
class cyclegan(object):
def __init__(self, opt):
self.discriminator = discriminator
self.generator = generator_resnet
self.criterionGAN = mae_criterion
self.opt = opt
self.img_shape = (self.opt.data_pix_size, self.opt.data_pix_size, self.opt.in_dim)
self.build_model()
self.pool = ImagePool(self.opt.max_size)
def build_model(self):
self.lambda_id = self.opt.lambda_id * self.opt.lambda_cycle
self.d_A = discriminator(self.opt)
self.d_B = discriminator(self.opt)
self.optimizer = Adam(self.opt.lr, self.opt.beta1)
self.d_A.compile(loss='mse',
optimizer=self.optimizer,
metrics=['accuracy'])
self.d_B.compile(loss='mse',
optimizer=self.optimizer,
metrics=['accuracy'])
self.g_AB = generator_resnet(self.opt)
self.g_BA = generator_resnet(self.opt)
img_A = Input(shape=self.img_shape)
img_B = Input(shape=self.img_shape)
#style transfer
fake_B = self.g_AB(img_A)
fake_A = self.g_BA(img_B)
#reconstructure
reconstr_A = self.g_BA(fake_B)
reconstr_B = self.g_AB(fake_A)
#identity
img_A_id = self.g_BA(img_A)
img_B_id = self.g_AB(img_B)
#to train generator
self.d_A.trainable = False
self.d_B.trainable = False
#dicriminator determine validity
valid_A = self.d_A(fake_A)
valid_B = self.d_B(fake_B)
# Combined model trains generators to fool discrgit/model.py:67iminators
self.combined = Model(inputs=[img_A, img_B],
outputs=[ valid_A, valid_B,
reconstr_A, reconstr_B,
img_A_id, img_B_id ])
self.combined.compile(loss=['mean_squared_error', 'mean_squared_error',
'mean_absolute_error', 'mean_absolute_error',
'mean_absolute_error', 'mean_absolute_error'],
loss_weights=[1, 1,
self.opt.lambda_cycle, self.opt.lambda_cycle,
self.lambda_id, self.lambda_id],
optimizer=self.optimizer)
def train(self):
start_time = datetime.datetime.now()
# Calculate output shape of D (PatchGAN)
patch = int(self.opt.d_patch_size)
self.disc_patch = (patch, patch, 1)
# Adversarial loss ground truths
valid = np.ones((self.opt.batch_size,) + self.disc_patch)
fake = np.zeros((self.opt.batch_size,) + self.disc_patch)
self.data_loader = DataLoader(dataset_name=self.opt.dataset_name,
img_res=(self.opt.data_pix_size, self.opt.data_pix_size))
for epoch in range(self.opt.epochs):
for batch_i, (imgs_A, imgs_B) in enumerate(self.data_loader.load_batch(self.opt.batch_size)):
# ----------------------
# Train Discriminators
# ----------------------
# Translate images to opposite domain
fake_B = self.g_AB.predict_on_batch(imgs_A)
fake_A = self.g_BA.predict_on_batch(imgs_B)
# Train the discriminators (original images = real / translated = Fake)
dA_loss_real = self.d_A.train_on_batch(imgs_A, valid)
dA_loss_fake = self.d_A.train_on_batch(fake_A, fake)
dA_loss = 0.5 * np.add(dA_loss_real, dA_loss_fake)
dB_loss_real = self.d_B.train_on_batch(imgs_B, valid)
dB_loss_fake = self.d_B.train_on_batch(fake_B, fake)
dB_loss = 0.5 * np.add(dB_loss_real, dB_loss_fake)
# Total disciminator loss
d_loss = 0.5 * np.add(dA_loss, dB_loss)
# ------------------
# Train Generators
# ------------------
# Train the generators
g_loss = self.combined.train_on_batch([imgs_A, imgs_B],
[valid, valid,
imgs_A, imgs_B,
imgs_A, imgs_B])
elapsed_time = datetime.datetime.now() - start_time
# Plot the progress
print ("[Epoch %d/%d] [Batch %d/%d] [D loss: %f, acc: %3d%%] [G loss: %05f, adv: %05f, recon: %05f, id: %05f] time: %s " \
% ( epoch, self.opt.epochs,
batch_i, self.data_loader.n_batches,
d_loss[0], 100*d_loss[1],
g_loss[0],
np.mean(g_loss[1:3]),
np.mean(g_loss[3:5]),
np.mean(g_loss[5:6]),
elapsed_time))
# If at save interval => save generated image samples
if batch_i % self.opt.sample_iter == 0:
self.sample_images(epoch, batch_i)
model_json = self.combined.to_json()
with open("model.json", "w") as json_file:
json_file.write(model_json)
# serialize weights to HDF5
self.combined.save_weights("model.h5")
print("Saved model to disk")
def sample_images(self,epoch, batch_i):
os.makedirs('images/%s' % self.opt.dataset_name, exist_ok=True)
r, c = 2, 3
imgs_A = self.data_loader.load_data(domain="A", batch_size=1, is_testing=True)
imgs_B = self.data_loader.load_data(domain="B", batch_size=1, is_testing=True)
# Demo (for GIF)
#imgs_A = self.data_loader.load_img('datasets/apple2orange/testA/n07740461_1541.jpg')
#imgs_B = self.data_loader.load_img('datasets/apple2orange/testB/n07749192_4241.jpg')
# Translate images to the other domain
fake_B = self.g_AB.predict(imgs_A)
fake_A = self.g_BA.predict(imgs_B)
# Translate back to original domain
reconstr_A = self.g_BA.predict(fake_B)
reconstr_B = self.g_AB.predict(fake_A)
gen_imgs = np.concatenate([imgs_A, fake_B, reconstr_A, imgs_B, fake_A, reconstr_B])
# Rescale images 0 - 1
gen_imgs = 0.5 * gen_imgs + 0.5
titles = ['Original', 'Translated', 'Reconstructed']
fig, axs = plt.subplots(r, c)
cnt = 0
for i in range(r):
for j in range(c):
axs[i,j].imshow(gen_imgs[cnt])
axs[i, j].set_title(titles[j])
axs[i,j].axis('off')
cnt += 1
fig.savefig("images/%s/%d_%d.png" % (self.opt.dataset_name, epoch, batch_i))
plt.close()
'''def test(self, opt):
"""Test cyclegan"""
init_op = tf.global_variables_initializer()
self.sess.run(init_op)
if opt.which_direction == 'AtoB':
sample_files = glob('./datasets/{}/*.*'.format(self.dataset_dir + '/testA'))
elif opt.which_direction == 'BtoA':
sample_files = glob('./datasets/{}/*.*'.format(self.dataset_dir + '/testB'))
else:
raise Exception('--which_direction must be AtoB or BtoA')
if self.load(opt.checkpoint_dir):
print(" [*] Load SUCCESS")
else:
print(" [!] Load failed...")
# write html for visual comparison
index_path = os.path.join(opt.test_dir, '{0}_index.html'.format(opt.which_direction))
index = open(index_path, "w")
index.write("<html><body><table><tr>")
index.write("<th>name</th><th>input</th><th>output</th></tr>")
out_var, in_var = (self.testB, self.test_A) if opt.which_direction == 'AtoB' else (
self.testA, self.test_B)
for sample_file in sample_files:
print('Processing image: ' + sample_file)
sample_image = [load_test_data(sample_file, opt.fine_size)]
sample_image = np.array(sample_image).astype(np.float32)
image_path = os.path.join(opt.test_dir,
'{0}_{1}'.format(opt.which_direction, os.path.basename(sample_file)))
fake_img = self.sess.run(out_var, feed_dict={in_var: sample_image})
save_images(fake_img, [1, 1], image_path)
index.write("<td>%s</td>" % os.path.basename(image_path))
index.write("<td><img src='%s'></td>" % (sample_file if os.path.isabs(sample_file) else (
'..' + os.path.sep + sample_file)))
index.write("<td><img src='%s'></td>" % (image_path if os.path.isabs(image_path) else (
'..' + os.path.sep + image_path)))
index.write("</tr>")
index.close()
'''
if __name__ == '__main__':
p = BaseOptions()
opt = p.args
gan = cyclegan(opt)
gan.train()