forked from CQFIO/PhotographicImageSynthesis
-
Notifications
You must be signed in to change notification settings - Fork 0
/
demo_1024p.py
148 lines (139 loc) · 8.41 KB
/
demo_1024p.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
from __future__ import division
import os,helper,time,scipy.io
import tensorflow as tf
import tensorflow.contrib.slim as slim
import numpy as np
def lrelu(x):
return tf.maximum(0.2*x,x)
def build_net(ntype,nin,nwb=None,name=None):
if ntype=='conv':
return tf.nn.relu(tf.nn.conv2d(nin,nwb[0],strides=[1,1,1,1],padding='SAME',name=name)+nwb[1])
elif ntype=='pool':
return tf.nn.avg_pool(nin,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')
def get_weight_bias(vgg_layers,i):
weights=vgg_layers[i][0][0][2][0][0]
weights=tf.constant(weights)
bias=vgg_layers[i][0][0][2][0][1]
bias=tf.constant(np.reshape(bias,(bias.size)))
return weights,bias
def build_vgg19(input,reuse=False):
if reuse:
tf.get_variable_scope().reuse_variables()
net={}
vgg_rawnet=scipy.io.loadmat('VGG_Model/imagenet-vgg-verydeep-19.mat')
vgg_layers=vgg_rawnet['layers'][0]
net['input']=input-np.array([123.6800, 116.7790, 103.9390]).reshape((1,1,1,3))
net['conv1_1']=build_net('conv',net['input'],get_weight_bias(vgg_layers,0),name='vgg_conv1_1')
net['conv1_2']=build_net('conv',net['conv1_1'],get_weight_bias(vgg_layers,2),name='vgg_conv1_2')
net['pool1']=build_net('pool',net['conv1_2'])
net['conv2_1']=build_net('conv',net['pool1'],get_weight_bias(vgg_layers,5),name='vgg_conv2_1')
net['conv2_2']=build_net('conv',net['conv2_1'],get_weight_bias(vgg_layers,7),name='vgg_conv2_2')
net['pool2']=build_net('pool',net['conv2_2'])
net['conv3_1']=build_net('conv',net['pool2'],get_weight_bias(vgg_layers,10),name='vgg_conv3_1')
net['conv3_2']=build_net('conv',net['conv3_1'],get_weight_bias(vgg_layers,12),name='vgg_conv3_2')
net['conv3_3']=build_net('conv',net['conv3_2'],get_weight_bias(vgg_layers,14),name='vgg_conv3_3')
net['conv3_4']=build_net('conv',net['conv3_3'],get_weight_bias(vgg_layers,16),name='vgg_conv3_4')
net['pool3']=build_net('pool',net['conv3_4'])
net['conv4_1']=build_net('conv',net['pool3'],get_weight_bias(vgg_layers,19),name='vgg_conv4_1')
net['conv4_2']=build_net('conv',net['conv4_1'],get_weight_bias(vgg_layers,21),name='vgg_conv4_2')
net['conv4_3']=build_net('conv',net['conv4_2'],get_weight_bias(vgg_layers,23),name='vgg_conv4_3')
net['conv4_4']=build_net('conv',net['conv4_3'],get_weight_bias(vgg_layers,25),name='vgg_conv4_4')
net['pool4']=build_net('pool',net['conv4_4'])
net['conv5_1']=build_net('conv',net['pool4'],get_weight_bias(vgg_layers,28),name='vgg_conv5_1')
net['conv5_2']=build_net('conv',net['conv5_1'],get_weight_bias(vgg_layers,30),name='vgg_conv5_2')
return net
def recursive_generator(label,sp):
dim=512 if sp>=128 else 1024
if sp==512:
dim=128
if sp==1024:
dim=32
if sp==4:
input=label
else:
downsampled=tf.image.resize_area(label,(sp//2,sp),align_corners=False)
input=tf.concat([tf.image.resize_bilinear(recursive_generator(downsampled,sp//2),(sp,sp*2),align_corners=True),label],3)
net=slim.conv2d(input,dim,[3,3],rate=1,normalizer_fn=slim.layer_norm,activation_fn=lrelu,scope='g_'+str(sp)+'_conv1')
net=slim.conv2d(net,dim,[3,3],rate=1,normalizer_fn=slim.layer_norm,activation_fn=lrelu,scope='g_'+str(sp)+'_conv2')
if sp==1024:
net=slim.conv2d(net,3,[1,1],rate=1,activation_fn=None,scope='g_'+str(sp)+'_conv100')
net=(net+1.0)/2.0*255.0
return net
def compute_error(real,fake,label):
#return tf.reduce_sum(tf.reduce_mean(label*tf.expand_dims(tf.reduce_mean(tf.abs(fake-real),reduction_indices=[3]),-1),reduction_indices=[1,2]))#diversity loss
return tf.reduce_mean(tf.abs(fake-real))#simple loss
#os.system('nvidia-smi -q -d Memory |grep -A4 GPU|grep Free >tmp')
#os.environ['CUDA_VISIBLE_DEVICES']=str(np.argmax([int(x.split()[2]) for x in open('tmp','r').readlines()]))#select a GPU with maximum available memory
#os.system('rm tmp')
sess=tf.Session()
is_training=False
sp=1024#spatial resolution: 1024x2048
with tf.variable_scope(tf.get_variable_scope()):
label=tf.placeholder(tf.float32,[None,None,None,20])
real_image=tf.placeholder(tf.float32,[None,None,None,3])
fake_image=tf.placeholder(tf.float32,[None,None,None,3])
generator=recursive_generator(label,sp)
weight=tf.placeholder(tf.float32)
vgg_real=build_vgg19(real_image)
vgg_fake=build_vgg19(generator,reuse=True)
p0=compute_error(vgg_real['input'],vgg_fake['input'],label)
p1=compute_error(vgg_real['conv1_2'],vgg_fake['conv1_2'],label)/2.6
p2=compute_error(vgg_real['conv2_2'],vgg_fake['conv2_2'],tf.image.resize_area(label,(sp//2,sp)))/4.8
p3=compute_error(vgg_real['conv3_2'],vgg_fake['conv3_2'],tf.image.resize_area(label,(sp//4,sp//2)))/3.7
p4=compute_error(vgg_real['conv4_2'],vgg_fake['conv4_2'],tf.image.resize_area(label,(sp//8,sp//4)))/5.6
p5=compute_error(vgg_real['conv5_2'],vgg_fake['conv5_2'],tf.image.resize_area(label,(sp//16,sp//8)))*10/1.5
G_loss=p0+p1+p2+p3+p4+p5
lr=tf.placeholder(tf.float32)
G_opt=tf.train.AdamOptimizer(learning_rate=lr).minimize(G_loss,var_list=[var for var in tf.trainable_variables() if var.name.startswith('g_1024') or var.name.startswith('g_512')])#only fine-tune the last two refinement module due to memory limitation
sess.run(tf.global_variables_initializer())
ckpt=tf.train.get_checkpoint_state("result_1024p")
if ckpt:
print('loaded '+ckpt.model_checkpoint_path)
saver=tf.train.Saver(var_list=[var for var in tf.trainable_variables() if var.name.startswith('g_')])
saver.restore(sess,ckpt.model_checkpoint_path)
else:
ckpt_prev=tf.train.get_checkpoint_state("result_512p")
saver=tf.train.Saver(var_list=[var for var in tf.trainable_variables() if var.name.startswith('g_') and not var.name.startswith('g_1024')])
print('loaded '+ckpt_prev.model_checkpoint_path)
saver.restore(sess,ckpt_prev.model_checkpoint_path)
saver=tf.train.Saver(max_to_keep=1000)
if is_training:
g_loss=np.zeros(3000,dtype=float)
input_images=[None]*3000
label_images=[None]*3000
for epoch in range(1,6):
if os.path.isdir("result_1024p/%04d"%epoch):
continue
cnt=0
for ind in np.random.permutation(2975)+1:
st=time.time()
cnt+=1
if input_images[ind] is None:
label_images[ind]=helper.get_semantic_map("data/cityscapes/Label1024Full/%08d.png"%ind)#training label
input_images[ind]=np.expand_dims(np.float32(scipy.misc.imread("data/cityscapes/RGB1024Full_vivid/%08d.png"%ind)),axis=0)#training image with vivid appearance. see "optional_preprocessing"
_,G_current,l0,l1,l2,l3,l4,l5=sess.run([G_opt,G_loss,p0,p1,p2,p3,p4,p5],feed_dict={label:np.concatenate((label_images[ind],np.expand_dims(1-np.sum(label_images[ind],axis=3),axis=3)),axis=3),real_image:input_images[ind],lr:1e-4})
g_loss[ind]=G_current
print("%d %d %.2f %.2f %.2f %.2f %.2f %.2f %.2f %.2f"%(epoch,cnt,np.mean(g_loss[np.where(g_loss)]),np.mean(l0),np.mean(l1),np.mean(l2),np.mean(l3),np.mean(l4),np.mean(l5),time.time()-st))
os.makedirs("result_1024p/%04d"%epoch)
target=open("result_1024p/%04d/score.txt"%epoch,'w')
target.write("%f"%np.mean(g_loss[np.where(g_loss)]))
target.close()
saver.save(sess,"result_1024p/model.ckpt")
if epoch%20==0:
saver.save(sess,"result_1024p/%04d/model.ckpt"%epoch)
for ind in range(100001,100051):
if not os.path.isfile("data/cityscapes/Label1024Full/%08d.png"%ind):#test label
continue
semantic=helper.get_semantic_map("data/cityscapes/Label1024Full/%08d.png"%ind)#test label
output=sess.run(generator,feed_dict={label:np.concatenate((semantic,np.expand_dims(1-np.sum(semantic,axis=3),axis=3)),axis=3)})
output=np.minimum(np.maximum(output,0.0),255.0)
scipy.misc.toimage(output[0,:,:,:],cmin=0,cmax=255).save("result_1024p/%04d/%06d_output.jpg"%(epoch,ind))
if not os.path.isdir("result_1024p/final"):
os.makedirs("result_1024p/final")
for ind in range(100001,100501):
if not os.path.isfile("data/cityscapes/Label1024Full/%08d.png"%ind):#test label
continue
semantic=helper.get_semantic_map("data/cityscapes/Label1024Full/%08d.png"%ind)#test label
output=sess.run(generator,feed_dict={label:np.concatenate((semantic,np.expand_dims(1-np.sum(semantic,axis=3),axis=3)),axis=3)})
output=np.minimum(np.maximum(output,0.0),255.0)
scipy.misc.toimage(output[0,:,:,:],cmin=0,cmax=255).save("result_1024p/final/%06d_output.jpg"%ind)