/
main.py
283 lines (198 loc) · 10.2 KB
/
main.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
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
import tensorflow as tf
import scipy.misc as sci
import os
import numpy as np
import Read_Image_List as ri
import module as mm
import ops as op
import random
import time
import cv2
Height = 256
Width = 256
batch_size = 8
mask_size = 128
dPath_l = ('./List')
dPath_train = ('/train_fh256.txt')
dPath_test = ('/test_fh256.txt')
dPath_testm = ('/test_mask256.txt')
dPath_testf = ('/test_maskff.txt')
name_f, num_f = ri.read_labeled_image_list(dPath_l + dPath_train)
name_test, num_test = ri.read_labeled_image_list(dPath_l + dPath_test)
name_testf, num_testf = ri.read_labeled_image_list(dPath_l + dPath_testf)
name_tests, num_tests, xst, yst = ri.read_labeled_image_list2(dPath_l + dPath_testm)
total_batch = int(num_f / batch_size)
save_path = './validation/v1'
model_path = './model/v1'
restore = False
restore_point = 900000
Checkpoint = model_path + '/cVG iter ' + str(restore_point) + '/'
WeightName = Checkpoint + 'Train_' + str(restore_point) + '.meta'
if restore == False:
restore_point = 0
saving_iter = 10000
Max_iter = 1000000
# ------- variables
X = tf.placeholder(tf.float32, [batch_size, Height, Width, 3])
Y = tf.placeholder(tf.float32, [batch_size, Height, Width, 3])
MASK = tf.placeholder(tf.float32, [batch_size, Height, Width, 3])
IT = tf.placeholder(tf.float32)
# ------- structure
input = tf.concat([X, MASK], 3)
vec_en = mm.encoder(input, reuse=False, name='G_en')
vec_con = mm.contextual_block(vec_en, vec_en, MASK, 3, 50.0, 'CB1', stride=1)
I_co = mm.decoder(vec_en, Height, reuse=False, name='G_de')
I_ge = mm.decoder(vec_con, Height, reuse=True, name='G_de')
image_result = I_ge * (1-MASK) + Y*MASK
D_real_red = mm.discriminator_red(Y, reuse=False, name='disc_red')
D_fake_red = mm.discriminator_red(image_result, reuse=True, name='disc_red')
# ------- Loss
Loss_D_red = tf.reduce_mean(tf.nn.relu(1+D_fake_red)) + tf.reduce_mean(tf.nn.relu(1-D_real_red))
Loss_D = Loss_D_red
Loss_gan_red = -tf.reduce_mean(D_fake_red)
Loss_gan = Loss_gan_red
Loss_s_re = tf.reduce_mean(tf.abs(I_ge - Y))
Loss_hat = tf.reduce_mean(tf.abs(I_co - Y))
A = tf.image.rgb_to_yuv((image_result+1)/2.0)
A_Y = tf.to_int32(A[:, :, :, 0:1]*255.0)
B = tf.image.rgb_to_yuv((Y+1)/2.0)
B_Y = tf.to_int32(B[:, :, :, 0:1]*255.0)
ssim = tf.reduce_mean(tf.image.ssim(A_Y, B_Y, 255.0))
alpha = IT/Max_iter
Loss_G = 0.1*Loss_gan + 10*Loss_s_re + 5*(1-alpha) * Loss_hat
# --------------------- variable & optimizer
var_D = [v for v in tf.global_variables() if v.name.startswith('disc_red')]
var_G = [v for v in tf.global_variables() if v.name.startswith('G_en') or v.name.startswith('G_de') or v.name.startswith('CB1')]
update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
with tf.control_dependencies(update_ops):
optimize_D = tf.train.AdamOptimizer(learning_rate=0.0004, beta1=0.5, beta2=0.9).minimize(Loss_D, var_list=var_D)
optimize_G = tf.train.AdamOptimizer(learning_rate=0.0001, beta1=0.5, beta2=0.9).minimize(Loss_G, var_list=var_G)
# --------- Run
config = tf.ConfigProto()
config.gpu_options.per_process_gpu_memory_fraction = 0.4
config.gpu_options.allow_growth = False
sess = tf.Session(config=config)
init = tf.global_variables_initializer()
sess.run(init)
saver = tf.train.Saver()
if restore == True:
print('Weight Restoring.....')
Restore = tf.train.import_meta_graph(WeightName)
Restore.restore(sess, tf.train.latest_checkpoint(Checkpoint))
print('Weight Restoring Finish!')
start_time = time.time()
for iter_count in range(restore_point, Max_iter + 1):
i = iter_count % total_batch
e = iter_count // total_batch
if i == 0:
np.random.shuffle(name_f)
data_g = ri.MakeImageBlock(name_f, Height, Width, i, batch_size)
data_temp = 255.0 * ((data_g + 1) / 2.0)
mask = op.ff_mask_batch(Height, batch_size, 50, 30, 3.14, 5, 15)
data_m = data_temp * mask
data_m = (data_m / 255.0) * 2.0 - 1
_, Loss1 = sess.run([optimize_D, Loss_D], feed_dict={X: data_m, Y: data_g, MASK: mask})
_, Loss2, Loss3 = sess.run([optimize_G, Loss_G, Loss_s_re], feed_dict={X: data_m, Y: data_g, MASK: mask, IT:iter_count})
if iter_count % 100 == 0:
consume_time = time.time() - start_time
print('%d Epoch : %d D Loss = %.5f G Loss = %.5f Recon Loss = %.5f time = %.4f' % (iter_count, e, Loss1, Loss2, Loss3, consume_time))
start_time = time.time()
if iter_count % saving_iter == 0:
print('SAVING MODEL')
Temp = model_path + '/cVG iter %s/' % iter_count
if not os.path.exists(Temp):
os.makedirs(Temp)
SaveName = (Temp + 'Train_%s' % (iter_count))
saver.save(sess, SaveName)
print('SAVING MODEL Finish')
psnr_l = 0
psnr_g = 0
psnr_f = 0
ssim_m = 0
num_s = random.sample(range(num_test - batch_size), 10)
for isave in range(5):
mask_sizet = random.randint(64, 128)
data_test = ri.MakeImageBlock(name_test, Height, Width, num_s[isave]//batch_size, batch_size)
data_tempt = 255.0 * ((data_test + 1) / 2.0)
mask_ts, xs, ys = op.make_sq_mask(Height, mask_sizet, batch_size)
mask_tf = op.ff_mask(Height, batch_size, 50, 20, 3.14, 6, 10)
data_tempts = data_tempt * mask_ts
data_mts = (data_tempts / 255.0) * 2.0 - 1
data_temptf = data_tempt * mask_tf
data_mtf = (data_temptf / 255.0) * 2.0 - 1
img_sample = sess.run(image_result, feed_dict={X: data_mts, Y: data_test, MASK: mask_ts})
img_sample2 = sess.run(image_result, feed_dict={X: data_mtf, Y: data_test, MASK: mask_tf})
for kk in range(batch_size):
temp_img1 = img_sample[kk,:,:,:]
temp_img2 = data_test[kk,:,:,:]
temp_img3 = img_sample2[kk, :, :, :]
img_gt = 255.0 * ((temp_img2 + 1) / 2.0)
img_ge = 255.0 * ((temp_img1 + 1) / 2.0)
img_ge2 = 255.0 * ((temp_img3 + 1) / 2.0)
Bigpaper1 = np.zeros((Height, 3 * Width + 60, 3))
Bigpaper1[0:Height, 0:Width, :] = img_gt
Bigpaper1[0:Height, Width + 30: 2 * Width + 30, :] = data_tempts[kk,:,:,:]
Bigpaper1[0: Height, 2 * Width + 60: 3 * Width + 60, :] = img_ge
Bigpaper2 = np.zeros((Height, 3 * Width + 60, 3))
Bigpaper2[0:Height, 0:Width, :] = img_gt
Bigpaper2[0:Height, Width + 30: 2 * Width + 30, :] = data_temptf[kk, :, :, :]
Bigpaper2[0: Height, 2 * Width + 60: 3 * Width + 60, :] = img_ge2
save_name = save_path + '/%04d' % iter_count
name = save_name + '/img_%02d_s.png' % (isave * batch_size + kk)
name2 = save_name + '/img_%02d_f.png' % (isave * batch_size + kk)
if not os.path.exists(save_name):
os.makedirs(save_name)
sci.imsave(name, Bigpaper1)
sci.imsave(name2, Bigpaper2)
for ipsnr in range(100):
mask_sizep = 128
data_test = ri.MakeImageBlock(name_test, Height, Width, ipsnr, batch_size)
data_tempt = 255.0 * ((data_test + 1) / 2.0)
mask_t = ri.MakeImageBlock(name_tests, Height, Width, ipsnr, batch_size)
mask_t = (mask_t + 1) / 2
data_tempt = data_tempt * mask_t
data_mt = (data_tempt / 255.0) * 2.0 - 1
img_sample1, ssim_temp = sess.run([image_result, ssim], feed_dict={X: data_mt, Y: data_test, MASK: mask_t})
for kk in range(batch_size):
xx = int(xst[ipsnr * batch_size + kk])
yy = int(yst[ipsnr * batch_size + kk])
img_sample2 = img_sample1[:, xx:xx + mask_sizep, yy:yy + mask_sizep, :]
img_sample3 = data_test[:, xx:xx + mask_sizep, yy:yy + mask_sizep, :]
temp_img1 = img_sample1[kk,:,:,:]
temp_img2 = img_sample2[kk,:,:,:]
temp_img3 = data_test[kk,:,:,:]
temp_img4 = img_sample3[kk,:,:,:]
img_re = 255.0 * ((temp_img1 + 1) / 2.0)
img_rem = 255.0 * ((temp_img2 + 1) / 2.0)
img_gt = 255.0 * ((temp_img3 + 1) / 2.0)
img_gtm = 255.0 * ((temp_img4 + 1) / 2.0)
mse_l = np.mean(np.square(img_gtm - img_rem))
mse_g = np.mean(np.square(img_gt - img_re))
psnr_l += 10 * np.log10(255.0 * 255.0 / mse_l)
psnr_g += 10 * np.log10(255.0 * 255.0 / mse_g)
ssim_m += ssim_temp
print('\nLocal = ', '%.4f' % (psnr_l/800),'\nGlobal = ', '%.4f\n' % (psnr_g/800), 'ssim = %.4f\n' % (ssim_m/100.0))
pp = open(save_path + '/PSNR_log.txt', 'a+')
data = '--------------------' + '\n%d' % iter_count + '\nLocal = ' + '%.4f' % (psnr_l / 800) + '\nGlobal = ' + '%.4f\n' % (psnr_g / 800) + 'ssim = %.4f\n' % (ssim_m/100.0)
pp.write(data)
pp.close()
for ipsnr in range(100):
data_test = ri.MakeImageBlock(name_test, Height, Width, ipsnr, batch_size)
data_tempt = 255.0 * ((data_test + 1) / 2.0)
mask_t = ri.MakeImageBlock(name_testf, Height, Width, ipsnr, batch_size)
mask_t = (mask_t + 1) / 2
data_tempt = data_tempt * mask_t
data_mt = (data_tempt / 255.0) * 2.0 - 1
img_sample1, ssim_temp = sess.run([image_result, ssim], feed_dict={X: data_mt, Y: data_test, MASK: mask_t})
for kk in range(batch_size):
temp_img1 = img_sample1[kk,:,:,:]
temp_img3 = data_test[kk,:,:,:]
img_re = 255.0 * ((temp_img1 + 1) / 2.0)
img_gt = 255.0 * ((temp_img3 + 1) / 2.0)
mse = np.mean(np.square(img_gt - img_re))
psnr_f += 10 * np.log10(255.0 * 255.0 / mse)
print('\nPSNR_f = ', '%.4f\n' % (psnr_f/800))
pp = open(save_path + '/PSNR_log.txt', 'a+')
data = '\nPSNR_f = ' + '%.4f\n' % (psnr_f / 800)
pp.write(data)
pp.close()