-
Notifications
You must be signed in to change notification settings - Fork 23
/
fast_scnn.py
182 lines (150 loc) · 7 KB
/
fast_scnn.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
import torch
from torch import nn
from torch.nn import functional as F
class Fast_SCNN(torch.nn.Module):
def __init__(self, input_channel, num_classes):
super().__init__()
self.learning_to_downsample = LearningToDownsample(input_channel)
self.global_feature_extractor = GlobalFeatureExtractor()
self.feature_fusion = FeatureFusionModule()
self.classifier = Classifier(num_classes)
def forward(self, x):
shared = self.learning_to_downsample(x)
x = self.global_feature_extractor(shared)
x = self.feature_fusion(shared, x)
x = self.classifier(x)
return x
class LearningToDownsample(torch.nn.Module):
def __init__(self, in_channels):
super().__init__()
self.conv1 = ConvBlock(in_channels=in_channels, out_channels=32, stride=2)
self.sconv1 = nn.Sequential(
nn.Conv2d(32, 32, kernel_size=3, stride=2, padding=1, dilation=1, groups=32, bias=False),
nn.BatchNorm2d(32),
nn.Conv2d(32, 48, kernel_size=1, stride=1, padding=0, dilation=1, groups=1, bias=False),
nn.BatchNorm2d(48),
nn.ReLU(inplace=True))
self.sconv2 = nn.Sequential(
nn.Conv2d(48, 48, kernel_size=3, stride=2, padding=1, dilation=1, groups=48, bias=False),
nn.BatchNorm2d(48),
nn.Conv2d(48, 64, kernel_size=1, stride=1, padding=0, dilation=1, groups=1, bias=False),
nn.BatchNorm2d(64),
nn.ReLU(inplace=True))
def forward(self, x):
x = self.conv1(x)
x = self.sconv1(x)
x = self.sconv2(x)
return x
class GlobalFeatureExtractor(torch.nn.Module):
def __init__(self):
super().__init__()
self.first_block = nn.Sequential(InvertedResidual(64, 64, 2, 6),
InvertedResidual(64, 64, 1, 6),
InvertedResidual(64, 64, 1, 6))
self.second_block = nn.Sequential(InvertedResidual(64, 96, 2, 6),
InvertedResidual(96, 96, 1, 6),
InvertedResidual(96, 96, 1, 6))
self.third_block = nn.Sequential(InvertedResidual(96, 128, 1, 6),
InvertedResidual(128, 128, 1, 6),
InvertedResidual(128, 128, 1, 6))
self.ppm = PSPModule(128, 128)
def forward(self, x):
x = self.first_block(x)
x = self.second_block(x)
x = self.third_block(x)
x = self.ppm(x)
return x
# Modified from https://github.com/tonylins/pytorch-mobilenet-v2
class InvertedResidual(nn.Module):
def __init__(self, inp, oup, stride, expand_ratio):
super(InvertedResidual, self).__init__()
self.stride = stride
assert stride in [1, 2]
hidden_dim = round(inp * expand_ratio)
self.use_res_connect = self.stride == 1 and inp == oup
if expand_ratio == 1:
self.conv = nn.Sequential(
# dw
nn.Conv2d(hidden_dim, hidden_dim, 3, stride, 1, groups=hidden_dim, bias=False),
nn.BatchNorm2d(hidden_dim),
nn.ReLU(inplace=True),
# pw-linear
nn.Conv2d(hidden_dim, oup, 1, 1, 0, bias=False),
nn.BatchNorm2d(oup),
)
else:
self.conv = nn.Sequential(
# pw
nn.Conv2d(inp, hidden_dim, 1, 1, 0, bias=False),
nn.BatchNorm2d(hidden_dim),
nn.ReLU(inplace=True),
# dw
nn.Conv2d(hidden_dim, hidden_dim, 3, stride, 1, groups=hidden_dim, bias=False),
nn.BatchNorm2d(hidden_dim),
nn.ReLU(inplace=True),
# pw-linear
nn.Conv2d(hidden_dim, oup, 1, 1, 0, bias=False),
nn.BatchNorm2d(oup),
)
def forward(self, x):
if self.use_res_connect:
return x + self.conv(x)
else:
return self.conv(x)
# Modified from https://github.com/Lextal/pspnet-pytorch/blob/master/pspnet.py
class PSPModule(nn.Module):
def __init__(self, features, out_features=1024, sizes=(1, 2, 3, 6)):
super().__init__()
self.stages = []
self.stages = nn.ModuleList([self._make_stage(features, size) for size in sizes])
self.bottleneck = nn.Conv2d(features * (len(sizes) + 1), out_features, kernel_size=1)
self.relu = nn.ReLU()
def _make_stage(self, features, size):
prior = nn.AdaptiveAvgPool2d(output_size=(size, size))
conv = nn.Conv2d(features, features, kernel_size=1, bias=False)
return nn.Sequential(prior, conv)
def forward(self, feats):
h, w = feats.size(2), feats.size(3)
priors = [F.interpolate(input=stage(feats), size=(h,w), mode='bilinear',
align_corners=True) for stage in self.stages] + [feats]
# import pdb;pdb.set_trace()
bottle = self.bottleneck(torch.cat(priors, 1))
return self.relu(bottle)
class FeatureFusionModule(torch.nn.Module):
def __init__(self):
super().__init__()
self.sconv1 = ConvBlock(in_channels=128, out_channels=128, stride=1, dilation=1, groups=128)
self.conv_low_res = nn.Conv2d(128, 128, kernel_size=1, stride=1, padding=0, bias=True)
self.conv_high_res = nn.Conv2d(64, 128, kernel_size=1, stride=1, padding=0, bias=True)
self.relu = nn.ReLU()
def forward(self, high_res_input, low_res_input):
low_res_input = F.interpolate(input=low_res_input, scale_factor=4, mode='bilinear', align_corners=True)
low_res_input = self.sconv1(low_res_input)
low_res_input = self.conv_low_res(low_res_input)
high_res_input = self.conv_high_res(high_res_input)
x = torch.add(high_res_input, low_res_input)
return self.relu(x)
class Classifier(torch.nn.Module):
def __init__(self, num_classes):
super().__init__()
self.sconv1 = ConvBlock(in_channels=128, out_channels=128, stride=1, dilation=1, groups=128)
self.sconv2 = ConvBlock(in_channels=128, out_channels=128, stride=1, dilation=1, groups=128)
self.conv = nn.Conv2d(128, num_classes, kernel_size=1, stride=1, padding=0, bias=True)
def forward(self, x):
x = self.sconv1(x)
x = self.sconv1(x)
return self.conv(x)
class ConvBlock(torch.nn.Module):
def __init__(self, in_channels, out_channels, kernel_size=3, stride=2, padding=1, dilation=1, groups=1):
super().__init__()
self.conv1 = nn.Conv2d(in_channels, out_channels, kernel_size=kernel_size, stride=stride, padding=padding,
dilation=dilation, groups=groups, bias=False)
self.bn = nn.BatchNorm2d(out_channels)
self.relu = nn.ReLU()
def forward(self, input):
x = self.conv1(input)
return self.relu(self.bn(x))
if __name__ == '__main__':
model = Fast_SCNN(input_channel=3, num_classes=10)
x = torch.rand(2, 3, 256, 256)
y = model(x)