-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathmodel.py
127 lines (117 loc) · 4.37 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
import torch
import torch.nn as nn
import math
import torch.nn.functional as F
'''
The codes of Residual Block is heavily borrowed from:
https://github.com/clovaai/stargan-v2/blob/master/core/model.py
'''
class ResBlk(nn.Module):
def __init__(self, dim_in, dim_out, actv=nn.LeakyReLU(0.2),
normalize=True, downsample=False):
super().__init__()
self.actv = actv
self.normalize = normalize
self.downsample = downsample
self.learned_sc = dim_in != dim_out
self._build_weights(dim_in, dim_out)
def _build_weights(self, dim_in, dim_out):
self.conv1 = nn.Conv2d(dim_in, dim_in, 3, 1, 1)
self.conv2 = nn.Conv2d(dim_in, dim_out, 3, 1, 1)
if self.normalize:
self.norm1 = nn.InstanceNorm2d(dim_in, affine=True)
self.norm2 = nn.InstanceNorm2d(dim_in, affine=True)
if self.learned_sc:
self.conv1x1 = nn.Conv2d(dim_in, dim_out, 1, 1, 0, bias=False)
def _shortcut(self, x):
if self.learned_sc:
x = self.conv1x1(x)
if self.downsample:
x = F.avg_pool2d(x, 2)
return x
def _residual(self, x):
if self.normalize:
x = self.norm1(x)
x = self.actv(x)
x = self.conv1(x)
if self.downsample:
x = F.avg_pool2d(x, 2)
if self.normalize:
x = self.norm2(x)
x = self.actv(x)
x = self.conv2(x)
return x
def forward(self, x):
x = self._shortcut(x) + self._residual(x)
return x / math.sqrt(2) # unit variance
class UpResBlk(nn.Module):
def _build_weights(self, dim_in, dim_out):
self.conv1 = nn.Conv2d(dim_in, dim_in, 3, 1, 1)
self.conv2 = nn.Conv2d(dim_in, dim_out, 3, 1, 1)
self.norm1 = nn.InstanceNorm2d(dim_in, affine=True)
self.norm2 = nn.InstanceNorm2d(dim_in, affine=True)
if self.learned_sc:
self.conv1x1 = nn.Conv2d(dim_in, dim_out, 1, 1, 0, bias=False)
def __init__(self, dim_in, dim_out, actv=nn.LeakyReLU(0.2)):
super().__init__()
self.learned_sc = dim_in != dim_out
self._build_weights(dim_in, dim_out)
self.actv = actv
def _shortcut(self, x):
x = F.interpolate(x, scale_factor=2, mode='nearest')
if self.learned_sc:
x = self.conv1x1(x)
return x
def _residual(self, x):
x = self.norm1(x)
x = self.actv(x)
x = F.interpolate(x, scale_factor=2, mode='nearest')
x = self.conv1(x)
x = self.norm2(x)
x = self.actv(x)
x = self.conv2(x)
return x
def forward(self, x):
out = self._residual(x)
out = (out + self._shortcut(x)) / math.sqrt(2)
return out
class Discriminator(nn.Module):
def btn_build(self):
btn = list()
for i in range(4):
btn.append(ResBlk(1024, 1024, downsample=False))
return nn.Sequential(*btn)
def __init__(self):
super().__init__()
self.encoder128 = ResBlk(3, 64, downsample=True)
self.encoder64 = ResBlk(64, 128, downsample=True)
self.encoder32 = ResBlk(128, 256, downsample=True)
self.encoder16 = ResBlk(256, 512, downsample=True)
self.encoder8 = ResBlk(512, 1024, downsample=True)
self.btn = self.btn_build()
self.decoder16 = UpResBlk(1024, 512)
self.decoder32 = UpResBlk(512, 256)
self.decoder64 = UpResBlk(256, 128)
self.decoder128 = UpResBlk(128, 64)
self.decoder256 = nn.Sequential(
UpResBlk(64, 1),
nn.Sigmoid()
)
def forward(self, x):
enc128 = self.encoder128(x)
enc64 = self.encoder64(enc128)
enc32 = self.encoder32(enc64)
enc16 = self.encoder16(enc32)
enc8 = self.encoder8(enc16)
enc8 = self.btn(enc8)
dec16 = self.decoder16(enc8)
dec32 = self.decoder32( dec16 * enc16 + enc16 )
dec64 = self.decoder64( dec32 * enc32 + enc32)
dec128 = self.decoder128(dec64 * enc64 + enc64)
dec256 = self.decoder256(dec128 * enc128 + enc128)
return dec256
if __name__ == "__main__":
toy_input = torch.randn((4, 3, 256, 256)).cuda()
discriminator = Discriminator().cuda()
p_map = discriminator(toy_input)
print(p_map.size())