-
Notifications
You must be signed in to change notification settings - Fork 28
/
Copy pathmodel.py
394 lines (323 loc) · 17.1 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
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
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
from __future__ import division
import os
import time
from glob import glob
import tensorflow as tf
import numpy as np
from six.moves import xrange
from ops import *
from utils import *
import pdb
# if use mnist, the y denotes the number in the image
# in generator, the z(noise) will concat with y as input
class CoGAN(object):
def __init__(self, sess, image_size=108, is_crop=True,
batch_size=64, sample_size = 64, output_size=64,
y_dim=None, z_dim=100, gf_dim=64, df_dim=64,
gfc_dim=1024, dfc_dim=1024, c_dim=3, dataset_name='default',
checkpoint_dir=None):
"""
Args:
sess: TensorFlow session
batch_size: The size of batch. Should be specified before training.
output_size: (optional) The resolution in pixels of the images. [64]
y_dim: (optional) Dimension of dim for y. [None]
z_dim: (optional) Dimension of dim for Z. [100]
gf_dim: (optional) Dimension of gen filters in first conv layer. [64]
df_dim: (optional) Dimension of discrim filters in first conv layer. [64]
gfc_dim: (optional) Dimension of gen units for for fully connected layer. [1024]
dfc_dim: (optional) Dimension of discrim units for fully connected layer. [1024]
c_dim: (optional) Dimension of image color. For grayscale input, set to 1. [3]
"""
self.sess = sess
self.is_crop = is_crop
self.is_grayscale = (c_dim == 1)
self.batch_size = batch_size
self.image_size = image_size
self.sample_size = sample_size
self.output_size = output_size
# y_dim is the conditional signal
self.y_dim = y_dim
self.z_dim = z_dim
self.gf_dim = gf_dim
self.df_dim = df_dim
self.gfc_dim = gfc_dim
self.dfc_dim = dfc_dim
self.c_dim = c_dim
# ------------batch norm-------------------
# batchnorm that share vars
self.g_bn0 = batch_norm(name='g_bn0')
self.g_bn1 = batch_norm(name='g_bn1')
self.g_bn2 = batch_norm(name='g_bn2')
self.d_bn2 = batch_norm(name='d_bn2')
# batchnorm that doesn't share vars
self.d1_bn1 = batch_norm(name='d1_bn1')
self.d2_bn1 = batch_norm(name='d2_bn1')
# -----------------------------------------
self.dataset_name = dataset_name
self.checkpoint_dir = checkpoint_dir
self.build_model()
def build_model(self):
# G1, D1
self.images1 = tf.placeholder(tf.float32, [self.batch_size] + [self.output_size, self.output_size, self.c_dim],
name='real_images1')
self.sample_images1 = tf.placeholder(tf.float32, [self.sample_size] + [self.output_size,self.output_size,self.c_dim],
name='sample_images1')
# G2, D2
self.images2 = tf.placeholder(tf.float32, [self.batch_size] + [self.output_size, self.output_size, self.c_dim],
name='real_images2')
self.sample_images2 = tf.placeholder(tf.float32, [self.sample_size] + [self.output_size,self.output_size,self.c_dim],
name='sample_images2')
# Generative model input
if self.y_dim:
self.y= tf.placeholder(tf.float32, [self.batch_size, self.y_dim], name='y')
# latent variable
self.z = tf.placeholder(tf.float32, [None, self.z_dim], name='z')
self.z_sum = tf.summary.histogram("z", self.z)
'''
The share_params FLAG denotes the share weight btn two network(G, D)
The reuse FLAG denotes that we will use it to do the inference
Note: every network should be declared with (False, False) FLAG
'''
# input of the generator is the concat of z, y
self.G1 = self.generator(self.z, self.y, share_params=False, reuse=False, name='G1')
self.G2 = self.generator(self.z, self.y, share_params=True, reuse=False, name='G2')
# input the paired input image(natural images)
self.D1_logits, self.D1 = self.discriminator(self.images1, self.y, share_params=False, reuse=False, name='D1')
self.D2_logits, self.D2 = self.discriminator(self.images2, self.y, share_params=True, reuse=False, name='D2')
# generate sample
self.sampler1 = self.generator(self.z, self.y, share_params=True, reuse=True, name='G1')
self.sampler2 = self.generator(self.z, self.y, share_params=True, reuse=True, name='G2')
# input the fake images
self.D1_logits_, self.D1_ = self.discriminator(self.G1, self.y, share_params=True, reuse=True, name='D1')
self.D2_logits_, self.D2_ = self.discriminator(self.G2, self.y, share_params=True, reuse=True, name='D2')
# B1
self.d1_sum = tf.summary.histogram("d1", self.D1)
self.d1__sum = tf.summary.histogram("d1_", self.D1_)
self.G1_sum = tf.summary.image("G1", self.G1)
# B2
self.d2_sum = tf.summary.histogram("d2", self.D2)
self.d2__sum = tf.summary.histogram("d2_", self.D2_)
self.G2_sum = tf.summary.image("G2", self.G2)
# B1
self.d1_loss_real = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(self.D1_logits, tf.ones_like(self.D1)*0.9))
self.d1_loss_fake = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(self.D1_logits_,tf.ones_like(self.D1_)*0.1))
self.g1_loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(self.D1_logits_, tf.ones_like(self.D1_)*0.9))
self.d1_loss_real_sum = tf.summary.scalar("d1_loss_real", self.d1_loss_real)
self.d1_loss_fake_sum = tf.summary.scalar("d1_loss_fake", self.d1_loss_fake)
self.d1_loss = self.d1_loss_real + self.d1_loss_fake
self.g1_loss_sum = tf.summary.scalar("g1_loss", self.g1_loss)
self.d1_loss_sum = tf.summary.scalar("d1_loss", self.d1_loss)
# B2
self.d2_loss_real = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(self.D2_logits, tf.ones_like(self.D2)*0.9))
self.d2_loss_fake = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(self.D2_logits_,tf.ones_like(self.D2_)*0.1))
self.g2_loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(self.D2_logits_, tf.ones_like(self.D2_)*0.9))
self.d2_loss_real_sum = tf.summary.scalar("d2_loss_real", self.d2_loss_real)
self.d2_loss_fake_sum = tf.summary.scalar("d2_loss_fake", self.d2_loss_fake)
self.d2_loss = self.d2_loss_real + self.d2_loss_fake
self.g2_loss_sum = tf.summary.scalar("g2_loss", self.g2_loss)
self.d2_loss_sum = tf.summary.scalar("d2_loss", self.d2_loss)
# sum together
self.d_loss = self.d1_loss+self.d2_loss
self.g_loss = self.g1_loss+self.g2_loss
self.g_loss_sum = tf.summary.scalar("g_loss", self.g_loss)
self.d_loss_sum = tf.summary.scalar("d_loss", self.d_loss)
# all variable
t_vars = tf.trainable_variables()
# variable list
self.g_vars = [var for var in t_vars if 'g1_' in var.name] + [var for var in t_vars if 'g2_' in var.name] \
+ [var for var in t_vars if 'g_' in var.name]
self.d_vars = [var for var in t_vars if 'd1_' in var.name] + [var for var in t_vars if 'd2_' in var.name] \
+ [var for var in t_vars if 'd_' in var.name]
self.saver = tf.train.Saver()
def train(self, config):
"""Train CoGAN"""
# data_X1 is the original image
# data_X2 is the black-white image
# data_y is the label
data_X1, data_y = self.load_mnist()
data_X2 = self.load_invert_mnist()
# do the random shuffle for two sets -> without paired images
idx = np.arange(len(data_y))
np.random.shuffle(idx)
data_X1 = data_X1[idx]
data_y1 = data_y[idx]
idx = np.arange(len(data_y))
np.random.shuffle(idx)
data_X2 = data_X2[idx]
data_y2 = data_y[idx]
d_optim = tf.train.AdamOptimizer(config.learning_rate, beta1=config.beta1) \
.minimize(self.d_loss, var_list=self.d_vars)
g_optim = tf.train.AdamOptimizer(config.learning_rate, beta1=config.beta1) \
.minimize(self.g_loss, var_list=self.g_vars)
tf.global_variables_initializer().run()
self.g1_sum = tf.summary.merge([self.z_sum, self.d1__sum,
self.G1_sum, self.d1_loss_fake_sum, self.g1_loss_sum])
self.d1_sum = tf.summary.merge([self.z_sum, self.d1_sum, self.d1_loss_real_sum, self.d1_loss_sum])
self.g2_sum = tf.summary.merge([self.z_sum, self.d2__sum,
self.G2_sum, self.d2_loss_fake_sum, self.g2_loss_sum])
self.d2_sum = tf.summary.merge([self.z_sum, self.d2_sum, self.d2_loss_real_sum, self.d2_loss_sum])
self.writer = tf.summary.FileWriter("./logs", self.sess.graph)
# sample noise
sample_z = np.random.normal(size=(self.batch_size , self.z_dim))
sample_images1 = data_X1[0:self.batch_size]
sample_images2 = data_X2[0:self.batch_size]
sample_labels1 = data_y1[0:self.batch_size]
sample_labels2 = data_y2[0:self.batch_size]
counter = 1
start_time = time.time()
if self.load(self.checkpoint_dir):
print(" [*] Load SUCCESS")
else:
print(" [!] Load failed...")
for epoch in xrange(config.epoch):
batch_idxs = min(len(data_X1), config.train_size) // config.batch_size
for idx in xrange(0, batch_idxs):
batch_images1 = data_X1[idx*config.batch_size:(idx+1)*config.batch_size]
batch_images2 = data_X2[idx*config.batch_size:(idx+1)*config.batch_size]
batch_labels1 = data_y1[idx*config.batch_size:(idx+1)*config.batch_size]
batch_labels2 = data_y2[idx*config.batch_size:(idx+1)*config.batch_size]
# z is the noise
batch_z = np.random.normal(size=[config.batch_size, self.z_dim]).astype(np.float32)
# Update D network
_, summary_str = self.sess.run([d_optim, self.d1_sum],
feed_dict={ self.images1: batch_images1, self.images2: batch_images2,
self.z: batch_z})
self.writer.add_summary(summary_str, counter)
# Update G network
_, summary_str = self.sess.run([g_optim, self.g1_sum],
feed_dict={ self.z: batch_z})
self.writer.add_summary(summary_str, counter)
# Run g_optim twice to make sure that d_loss does not go to zero (different from paper)
_, summary_str = self.sess.run([g_optim, self.g1_sum],
feed_dict={ self.z: batch_z})
self.writer.add_summary(summary_str, counter)
errD = self.d_loss.eval({self.z: batch_z, self.images1: batch_images1, self.images2: batch_images2})
errG = self.g_loss.eval({self.z: batch_z})
counter += 1
print("Epoch: [%2d] [%4d/%4d] time: %4.4f, d_loss: %.8f, g_loss: %.8f" \
% (epoch, idx, batch_idxs,
time.time() - start_time, errD, errG))
if np.mod(counter, 100) == 1:
self.evaluate(sample_images1,sample_images2,sample_labels1,batch_labels1,sample_labels2,batch_labels2,
sample_z, './samples/top/train_{:02d}_{:04d}.png'.format(epoch, idx))
if np.mod(counter, 500) == 2:
self.save(config.checkpoint_dir, counter)
def evaluate(self, sample_images1=None, sample_images2=None, sample_labels1=None,batch_labels1=None,
sample_labels2=None,batch_labels2=None, sample_z=None, img_name=None):
if sample_images1==None:
data_X1, data_y = self.load_mnist()
data_X2 = self.load_invert_mnist()
# sample noise
sample_z = np.random.uniform(-1, 1, size=(self.batch_size , self.z_dim))
sample_images1 = data_X1[0:self.batch_size]
sample_images2 = data_X2[0:self.batch_size]
sample_labels = data_y[0:self.batch_size]
img_name = './evaluate/top/testing'
samples1, d1_loss, g1_loss = self.sess.run(
[self.sampler1, self.d1_loss, self.g1_loss],
feed_dict={self.z: sample_z, self.images1: sample_images1, self.y:batch_labels1}
)
save_images(samples1[:self.sample_size], [8, 8], img_name, type='BW')
print("[Sample T] d_loss: %.8f, g_loss: %.8f" % (d1_loss, g1_loss))
# sample is the generated image
samples2, d2_loss, g2_loss = self.sess.run(
[self.sampler2, self.d2_loss, self.g2_loss],
feed_dict={self.z: sample_z, self.images2: sample_images2, self.y:batch_labels2}
)
save_images(samples2[:self.sample_size], [8, 8], img_name.replace('top', 'bot'), type='BW')
print("[Sample B] d_loss: %.8f, g_loss: %.8f" % (d2_loss, g2_loss))
def discriminator(self, image, y=None, share_params=False, reuse=False, name='D'):
# select the corresponding batchnorm1(not shared)
if '1' in name:
d_bn1 = self.d1_bn1
branch = '1'
elif '2' in name:
d_bn1 = self.d2_bn1
branch = '2'
# layers that don't share variable
with tf.variable_scope(name):
if reuse:
tf.get_variable_scope().reuse_variables()
h0 = prelu(conv2d(image, self.c_dim, name='d'+branch+'_h0_conv', reuse=False),
name='d'+branch+'_h0_prelu', reuse=False)
h1 = prelu(d_bn1(conv2d(h0, self.df_dim, name='d'+branch+'_h1_conv', reuse=False), reuse=reuse),
name='d'+branch+'_h1_prelu', reuse=False)
h1 = tf.reshape(h1, [self.batch_size, -1])
# layers that share variables
h2 = prelu(self.d_bn2(linear(h1, self.dfc_dim, 'd_h2_lin', reuse=share_params),reuse=share_params),
name='d_h2_prelu', reuse=share_params)
h3 = linear(h2, 1, 'd_h3_lin', reuse=share_params)
return tf.nn.sigmoid(h3), h3
def generator(self, z, y=None, share_params=False, reuse=False, name='G'):
if '1' in name:
branch = '1'
elif '2' in name:
branch = '2'
# layers that share the variables
s = self.output_size
s2, s4 = int(s/2), int(s/4)
h0 = prelu(self.g_bn0(linear(z, self.gfc_dim, 'g_h0_lin', reuse=share_params), reuse=share_params),
name='g_h0_prelu', reuse=share_params)
h1 = prelu(self.g_bn1(linear(z, self.gf_dim*2*s4*s4,'g_h1_lin',reuse=share_params),reuse=share_params),
name='g_h1_prelu', reuse=share_params)
h1 = tf.reshape(h1, [self.batch_size, s4, s4, self.gf_dim * 2])
h2 = prelu(self.g_bn2(deconv2d(h1, [self.batch_size,s2,s2,self.gf_dim * 2],
name='g_h2', reuse=share_params), reuse=share_params), name='g_h2_prelu', reuse=share_params)
# layers that don't share the variable
with tf.variable_scope(name):
if reuse:
tf.get_variable_scope().reuse_variables()
output = tf.nn.sigmoid(deconv2d(h2, [self.batch_size, s, s, self.c_dim], name='g'+branch+'_h3', reuse=False))
return output
def load_invert_mnist(self):
data_dir = os.path.join(os.path.join("./data", self.dataset_name, 'invert'))
trX = np.load(os.path.join(data_dir, 'train-images-idx3-ubyte.npy'))
teX = np.load(os.path.join(data_dir, 't10k-images-idx3-ubyte.npy'))
X = np.concatenate((trX, teX), axis=0)
# convert to 0~1 is more convenient
return X/255.
def load_mnist(self):
data_dir = os.path.join("./data", self.dataset_name)
fd = open(os.path.join(data_dir,'train-images-idx3-ubyte'))
loaded = np.fromfile(file=fd,dtype=np.uint8)
trX = loaded[16:].reshape((60000,28,28,1)).astype(np.float)
fd = open(os.path.join(data_dir,'train-labels-idx1-ubyte'))
loaded = np.fromfile(file=fd,dtype=np.uint8)
trY = loaded[8:].reshape((60000)).astype(np.float)
fd = open(os.path.join(data_dir,'t10k-images-idx3-ubyte'))
loaded = np.fromfile(file=fd,dtype=np.uint8)
teX = loaded[16:].reshape((10000,28,28,1)).astype(np.float)
fd = open(os.path.join(data_dir,'t10k-labels-idx1-ubyte'))
loaded = np.fromfile(file=fd,dtype=np.uint8)
teY = loaded[8:].reshape((10000)).astype(np.float)
trY = np.asarray(trY)
teY = np.asarray(teY)
X = np.concatenate((trX, teX), axis=0)
y = np.concatenate((trY, teY), axis=0)
# convert label into one-hot
y_vec = np.zeros((len(y), self.y_dim), dtype=np.float)
for i, label in enumerate(y):
y_vec[i,int(y[i])] = 1.0
# conver to 0~1 is more convenient
return X/255.,y_vec
def save(self, checkpoint_dir, step):
model_name = "CoGAN.model"
model_dir = "%s_%s_%s" % (self.dataset_name, self.batch_size, self.output_size)
checkpoint_dir = os.path.join(checkpoint_dir, model_dir)
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
self.saver.save(self.sess,
os.path.join(checkpoint_dir, model_name),
global_step=step)
def load(self, checkpoint_dir):
print(" [*] Reading checkpoints...")
model_dir = "%s_%s_%s" % (self.dataset_name, self.batch_size, self.output_size)
checkpoint_dir = os.path.join(checkpoint_dir, model_dir)
ckpt = tf.train.get_checkpoint_state(checkpoint_dir)
if ckpt and ckpt.model_checkpoint_path:
ckpt_name = os.path.basename(ckpt.model_checkpoint_path)
self.saver.restore(self.sess, os.path.join(checkpoint_dir, ckpt_name))
return True
else:
return False