-
Notifications
You must be signed in to change notification settings - Fork 27
/
main_3D-RecGAN++.py
282 lines (241 loc) · 12.6 KB
/
main_3D-RecGAN++.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
import os
import shutil
import numpy as np
import scipy.io
import tensorflow as tf
import tools
vox_res64 = 64
vox_rex256 = 256
batch_size = 4
GPU0 = '0'
re_train=False
#########################
config={}
config['batch_size']=batch_size
config['vox_res_x'] = vox_res64
config['vox_res_y'] = vox_rex256
config['train_names']=['P1_02828884_bench','P1_03001627_chair','P1_04256520_couch', 'P1_04379243_table']
for name in config['train_names']:
config['X_train_'+name] = './Data_sample/'+name+'/train_125_25d_vox256/'
config['Y_train_'+name] = './Data_sample/'+name+'/train_125_3d_vox256/'
config['test_names']=['P1_02828884_bench','P1_03001627_chair','P1_04256520_couch', 'P1_04379243_table']
for name in config['test_names']:
config['X_test_'+name]= './Data_sample/'+name+'/test_125_25d_vox256/'
config['Y_test_'+name]= './Data_sample/'+name+'/test_125_3d_vox256/'
#########################
class Network:
def __init__(self, demo_only=False):
if demo_only:
return # no need to creat folders
self.train_mod_dir = './train_mod/'
self.train_sum_dir = './train_sum/'
self.test_res_dir = './test_res/'
self.test_sum_dir = './test_sum/'
print ("re_train:", re_train)
if os.path.exists(self.test_res_dir):
if re_train:
print ("test_res_dir and files kept!")
else:
shutil.rmtree(self.test_res_dir)
os.makedirs(self.test_res_dir)
print ('test_res_dir: deleted and then created!')
else:
os.makedirs(self.test_res_dir)
print ('test_res_dir: created!')
if os.path.exists(self.train_mod_dir):
if re_train:
if os.path.exists(self.train_mod_dir + 'model.cptk.data-00000-of-00001'):
print ('model found! will be reused!')
else:
print ('model not found! error!')
exit()
else:
shutil.rmtree(self.train_mod_dir)
os.makedirs(self.train_mod_dir)
print ('train_mod_dir: deleted and then created!')
else:
os.makedirs(self.train_mod_dir)
print ('train_mod_dir: created!')
if os.path.exists(self.train_sum_dir):
if re_train:
print ("train_sum_dir and files kept!")
else:
shutil.rmtree(self.train_sum_dir)
os.makedirs(self.train_sum_dir)
print ('train_sum_dir: deleted and then created!')
else:
os.makedirs(self.train_sum_dir)
print ('train_sum_dir: created!')
if os.path.exists(self.test_sum_dir):
if re_train:
print ("test_sum_dir and files kept!")
else:
shutil.rmtree(self.test_sum_dir)
os.makedirs(self.test_sum_dir)
print ('test_sum_dir: deleted and then created!')
else:
os.makedirs(self.test_sum_dir)
print ('test_sum_dir: created!')
def aeu(self, X):
with tf.device('/gpu:'+GPU0):
X = tf.reshape(X,[-1, vox_res64,vox_res64,vox_res64,1])
c_e = [1,64,128,256,512]
s_e = [0,1 , 1, 1, 1]
layers_e = []
layers_e.append(X)
for i in range(1,5,1):
layer = tools.Ops.conv3d(layers_e[-1],k=4,out_c=c_e[i],str=s_e[i],name='e'+str(i))
layer = tools.Ops.maxpool3d(tools.Ops.xxlu(layer, label='lrelu'), k=2,s=2,pad='SAME')
layers_e.append(layer)
### fc
[_, d1, d2, d3, cc] = layers_e[-1].get_shape()
d1=int(d1); d2=int(d2); d3=int(d3); cc=int(cc)
lfc = tf.reshape(layers_e[-1],[-1, int(d1)*int(d2)*int(d3)*int(cc)])
lfc = tools.Ops.xxlu(tools.Ops.fc(lfc, out_d=2000,name='fc1'), label='relu')
with tf.device('/gpu:'+GPU0):
lfc = tools.Ops.xxlu(tools.Ops.fc(lfc,out_d=d1*d2*d3*cc, name='fc2'), label='relu')
lfc = tf.reshape(lfc, [-1, d1,d2,d3,cc])
c_d = [0,256,128,64,16,8]
s_d = [0,2,2,2,2,2]
layers_d = []
layers_d.append(lfc)
for j in range(1,6,1):
u_net = True
if u_net:
layer = tf.concat([layers_d[-1], layers_e[-j]],axis=4)
layer = tools.Ops.deconv3d(layer, k=4,out_c=c_d[j], str=s_d[j],name='d'+str(len(layers_d)))
else:
layer = tools.Ops.deconv3d(layers_d[-1],k=4,out_c=c_d[j],str=s_d[j],name='d'+str(len(layers_d)))
layer = tools.Ops.xxlu(layer, label='relu')
layers_d.append(layer)
###
layer = tools.Ops.deconv3d(layers_d[-1],k=4,out_c=1,str=2,name='dlast')
###
Y_sig = tf.nn.sigmoid(layer)
Y_sig_modi = tf.maximum(Y_sig,0.01)
return Y_sig, Y_sig_modi
def dis(self, X, Y):
with tf.device('/gpu:'+GPU0):
X = tf.reshape(X,[-1, vox_res64, vox_res64, vox_res64,1])
X = tf.reshape(X, [-1, vox_rex256, vox_rex256, 4, 1])
Y = tf.reshape(Y,[-1, vox_rex256, vox_rex256,vox_rex256,1])
Y = tf.concat([X, Y],axis=3)
c_d = [1,8,16,32,64,128,256]
s_d = [0,2,2,2,2,2,2]
layers_d =[]
layers_d.append(Y)
for i in range(1,7,1):
layer = tools.Ops.conv3d(layers_d[-1],k=4,out_c=c_d[i],str=s_d[i],name='d'+str(i))
if i!=6:
layer = tools.Ops.xxlu(layer, label='lrelu')
layers_d.append(layer)
[_, d1, d2, d3, cc] = layers_d[-1].get_shape()
d1 = int(d1); d2 = int(d2); d3 = int(d3); cc = int(cc)
y = tf.reshape(layers_d[-1],[-1,d1*d2*d3*cc])
return tf.nn.sigmoid(y)
def build_graph(self):
self.X = tf.placeholder(shape=[None, vox_res64, vox_res64, vox_res64, 1], dtype=tf.float32)
self.Y = tf.placeholder(shape=[None, vox_rex256, vox_rex256, vox_rex256, 1], dtype=tf.float32)
with tf.variable_scope('aeu'):
self.Y_pred, self.Y_pred_modi = self.aeu(self.X)
with tf.variable_scope('dis'):
self.XY_real_pair = self.dis(self.X, self.Y)
with tf.variable_scope('dis',reuse=True):
self.XY_fake_pair = self.dis(self.X, self.Y_pred)
with tf.device('/gpu:'+GPU0):
################################ ae loss
Y_ = tf.reshape(self.Y, shape=[-1, vox_rex256**3])
Y_pred_modi_ = tf.reshape(self.Y_pred_modi, shape=[-1, vox_rex256**3])
w = 0.85
self.aeu_loss = tf.reduce_mean(-tf.reduce_mean(w * Y_ * tf.log(Y_pred_modi_ + 1e-8), reduction_indices=[1]) -
tf.reduce_mean((1 - w) * (1 - Y_) * tf.log(1 - Y_pred_modi_ + 1e-8), reduction_indices=[1]))
sum_aeu_loss = tf.summary.scalar('aeu_loss', self.aeu_loss)
################################ wgan loss
self.gan_g_loss = -tf.reduce_mean(self.XY_fake_pair)
self.gan_d_loss_no_gp = tf.reduce_mean(self.XY_fake_pair) - tf.reduce_mean(self.XY_real_pair)
sum_gan_g_loss = tf.summary.scalar('gan_g_loss', self.gan_g_loss)
sum_gan_d_loss_no_gp = tf.summary.scalar('gan_d_loss_no_gp', self.gan_d_loss_no_gp)
alpha = tf.random_uniform(shape=[tf.shape(self.X)[0], vox_rex256 ** 3], minval=0.0, maxval=1.0)
Y_pred_ = tf.reshape(self.Y_pred, shape=[-1, vox_rex256 ** 3])
differences_ = Y_pred_ - Y_
interpolates = Y_ + alpha*differences_
with tf.variable_scope('dis',reuse=True):
XY_fake_intep = self.dis(self.X, interpolates)
gradients = tf.gradients(XY_fake_intep, [interpolates])[0]
slopes = tf.sqrt(tf.reduce_sum(tf.square(gradients), reduction_indices=[1]))
gradient_penalty = tf.reduce_mean((slopes - 1.0) ** 2)
self.gan_d_loss_gp = self.gan_d_loss_no_gp + 10 * gradient_penalty
sum_gan_d_loss_gp = tf.summary.scalar('gan_d_loss_gp', self.gan_d_loss_gp)
################################# ae + gan loss
gan_g_w = 20
aeu_w = 100 - gan_g_w
self.aeu_gan_g_loss = aeu_w*self.aeu_loss + gan_g_w*self.gan_g_loss
with tf.device('/gpu:'+GPU0):
aeu_var = [var for var in tf.trainable_variables() if var.name.startswith('aeu')]
dis_var = [var for var in tf.trainable_variables() if var.name.startswith('dis')]
self.aeu_g_optim = tf.train.AdamOptimizer(learning_rate=0.0001, beta1=0.9, beta2=0.999, epsilon=1e-8).\
minimize(self.aeu_gan_g_loss, var_list=aeu_var)
self.dis_optim = tf.train.AdamOptimizer(learning_rate=0.00005, beta1=0.9, beta2=0.999, epsilon=1e-8).\
minimize(self.gan_d_loss_gp,var_list=dis_var)
print (tools.Ops.variable_count())
self.sum_merged = tf.summary.merge_all()
self.saver = tf.train.Saver(max_to_keep=1)
config = tf.ConfigProto(allow_soft_placement=True)
config.gpu_options.visible_device_list = GPU0
self.sess = tf.Session(config=config)
self.sum_writer_train = tf.summary.FileWriter(self.train_sum_dir, self.sess.graph)
self.sum_write_test = tf.summary.FileWriter(self.test_sum_dir)
path = self.train_mod_dir
#path = './Model_released/' # to retrain our released model
if os.path.isfile(path + 'model.cptk.data-00000-of-00001'):
print ('restoring saved model')
self.saver.restore(self.sess, path + 'model.cptk')
else:
print ('initilizing model')
self.sess.run(tf.global_variables_initializer())
return 0
def train(self, data):
for epoch in range(10):
data.shuffle_X_Y_files(label='train')
total_train_batch_num = data.total_train_batch_num
print ('total_train_batch_num:', total_train_batch_num)
for i in range(total_train_batch_num):
#################### training
X_train_batch, Y_train_batch = data.queue_train.get()
self.sess.run(self.dis_optim, feed_dict={self.X:X_train_batch, self.Y:Y_train_batch})
self.sess.run(self.aeu_g_optim, feed_dict={self.X:X_train_batch, self.Y:Y_train_batch})
aeu_loss_c, gan_g_loss_c, gan_d_loss_no_gp_c, gan_d_loss_gp_c, sum_train = self.sess.run(
[self.aeu_loss, self.gan_g_loss, self.gan_d_loss_no_gp, self.gan_d_loss_gp, self.sum_merged],
feed_dict={self.X:X_train_batch, self.Y:Y_train_batch})
if i%200==0:
self.sum_writer_train.add_summary(sum_train, epoch * total_train_batch_num + i)
print ('ep:',epoch,'i:',i, 'train aeu loss:',aeu_loss_c, 'gan g loss:',gan_g_loss_c,
'gan d loss no gp:',gan_d_loss_no_gp_c,'gan d loss gp:', gan_d_loss_gp_c)
#################### testing
if i%600==0:
X_test_batch, Y_test_batch = data.load_X_Y_voxel_grids_test_next_batch()
aeu_loss_t, gan_g_loss_t, gan_d_loss_no_gp_t, gan_d_loss_gp_t, Y_pred_t, sum_test = self.sess.run(
[self.aeu_loss, self.gan_g_loss, self.gan_d_loss_no_gp, self.gan_d_loss_gp, self.Y_pred, self.sum_merged],
feed_dict={self.X:X_test_batch, self.Y:Y_test_batch})
X_test_batch=X_test_batch.astype(np.int8)
Y_pred_t=Y_pred_t.astype(np.float16)
Y_test_batch=Y_test_batch.astype(np.int8)
to_save = {'X_test':X_test_batch, 'Y_test_pred':Y_pred_t, 'Y_test_true':Y_test_batch}
scipy.io.savemat(self.test_res_dir+'X_Y_pred_'+str(epoch).zfill(2)+'_'+str(i).zfill(5)+'.mat',
to_save, do_compression=True)
self.sum_write_test.add_summary(sum_test, epoch*total_train_batch_num+i)
print ('ep:',epoch, 'i:', i, 'test aeu loss:', aeu_loss_t,'gan g loss:', gan_g_loss_t,
'gan d loss no gp:',gan_d_loss_no_gp_t,'gan d loss gp:',gan_d_loss_gp_t)
#### model saving
if i%600 == 0 and i > 0:
self.saver.save(self.sess, save_path=self.train_mod_dir + 'model.cptk')
print ('ep:', epoch, 'i:', i, 'model saved!')
data.stop_queue=True
#########################
if __name__ == '__main__':
data = tools.Data(config)
data.daemon = True
data.start()
net = Network()
net.build_graph()
net.train(data)