-
Notifications
You must be signed in to change notification settings - Fork 4
/
erfpspnet_hier42.py
228 lines (165 loc) · 7.43 KB
/
erfpspnet_hier42.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
import torch
import torch.nn as nn
import torch.nn.init as init
import torch.nn.functional as F
class DownsamplerBlock (nn.Module):
def __init__(self, ninput, noutput):
super().__init__()
self.conv = nn.Conv2d(ninput, noutput-ninput, (3, 3), stride=2, padding=1, bias=True)
self.pool = nn.MaxPool2d(2, stride=2)
self.bn = nn.BatchNorm2d(noutput, eps=1e-3)
def forward(self, input):
output = torch.cat([self.conv(input), self.pool(input)], 1)
output = self.bn(output)
return F.relu(output)
class non_bottleneck_1d (nn.Module):
def __init__(self, chann, dropprob, dilated):
super().__init__()
self.conv3x1_1 = nn.Conv2d(chann, chann, (3, 1), stride=1, padding=(1,0), bias=True)
self.conv1x3_1 = nn.Conv2d(chann, chann, (1,3), stride=1, padding=(0,1), bias=True)
self.bn1 = nn.BatchNorm2d(chann, eps=1e-03)
self.conv3x1_2 = nn.Conv2d(chann, chann, (3, 1), stride=1, padding=(1*dilated,0), bias=True, dilation = (dilated,1))
self.conv1x3_2 = nn.Conv2d(chann, chann, (1,3), stride=1, padding=(0,1*dilated), bias=True, dilation = (1, dilated))
self.bn2 = nn.BatchNorm2d(chann, eps=1e-03)
self.dropout = nn.Dropout2d(dropprob)
def forward(self, input):
output = self.conv3x1_1(input)
output = F.relu(output)
output = self.conv1x3_1(output)
output = self.bn1(output)
output = F.relu(output)
output = self.conv3x1_2(output)
output = F.relu(output)
output = self.conv1x3_2(output)
output = self.bn2(output)
#output = F.relu(output)
if (self.dropout.p != 0):
output = self.dropout(output)
return F.relu(output+input) #+input = identity (residual connection)
class non_bottleneck_1d_hier (nn.Module):
def __init__(self):
super().__init__()
self.conv3x1_1 = nn.Conv2d(128, 128, (3, 1), stride=1, padding=(1,0), bias=True)
self.conv1x3_1 = nn.Conv2d(128, 128, (1,3), stride=1, padding=(0,1), bias=True)
self.bn1 = nn.BatchNorm2d(128, eps=1e-03)
self.conv3x1_22 = nn.Conv2d(128, 128, (3, 1), stride=1, padding=(2,0), bias=True, dilation = (2,1))
self.conv1x3_22 = nn.Conv2d(128, 128, (1,3), stride=1, padding=(0,2), bias=True, dilation = (1, 2))
self.conv3x1_24 = nn.Conv2d(128, 128, (3, 1), stride=1, padding=(4,0), bias=True, dilation = (4,1))
self.conv1x3_24 = nn.Conv2d(128, 128, (1,3), stride=1, padding=(0,4), bias=True, dilation = (1, 4))
self.conv3x1_28 = nn.Conv2d(128, 128, (3, 1), stride=1, padding=(8,0), bias=True, dilation = (8,1))
self.conv1x3_28 = nn.Conv2d(128, 128, (1,3), stride=1, padding=(0,8), bias=True, dilation = (1, 8))
self.conv3x1_216 = nn.Conv2d(128, 128, (3, 1), stride=1, padding=(16,0), bias=True, dilation = (16,1))
self.conv1x3_216 = nn.Conv2d(128, 128, (1,3), stride=1, padding=(0,16), bias=True, dilation = (1, 16))
self.bn2 = nn.BatchNorm2d(128, eps=1e-03)
self.dropout = nn.Dropout2d(0.3)
def forward(self, input):
output = self.conv3x1_1(input)
output = F.relu(output)
output = self.conv1x3_1(output)
output = self.bn1(output)
output = F.relu(output)
output2 = self.conv3x1_22(output)
output2 = F.relu(output2)
output2 = self.conv1x3_22(output2)
output2 = self.bn2(output2)
if (self.dropout.p != 0):
output2 = self.dropout(output2)
output4 = self.conv3x1_24(output)
output4 = F.relu(output4)
output4 = self.conv1x3_24(output4)
output4 = self.bn2(output4)
if (self.dropout.p != 0):
output4 = self.dropout(output4)
output8 = self.conv3x1_28(output)
output8 = F.relu(output8)
output8 = self.conv1x3_28(output8)
output8 = self.bn2(output8)
if (self.dropout.p != 0):
output8 = self.dropout(output8)
output16 = self.conv3x1_216(output)
output16 = F.relu(output16)
output16 = self.conv1x3_216(output16)
output16 = self.bn2(output16)
if (self.dropout.p != 0):
output16 = self.dropout(output16)
return F.relu(output2+output4+output8+output16+input)
class Encoder(nn.Module):
def __init__(self, num_classes):
super().__init__()
self.initial_block = DownsamplerBlock(3,16)
self.layers = nn.ModuleList()
self.layers.append(DownsamplerBlock(16,64))
for x in range(0, 5): #5 times
self.layers.append(non_bottleneck_1d(64, 0.03, 1))
self.layers.append(DownsamplerBlock(64,128))
for x in range(0, 2): #2 times
self.layers.append(non_bottleneck_1d_hier())
self.output_conv = nn.Conv2d(128, num_classes, 1, stride=1, padding=0, bias=True)
def forward(self, input, predict=False):
output = self.initial_block(input)
for layer in self.layers:
output = layer(output)
if predict:
output = self.output_conv(output)
return output
class UpsamplerBlock (nn.Module):
def __init__(self, ninput, noutput):
super().__init__()
self.conv = nn.ConvTranspose2d(ninput, noutput, 3, stride=2, padding=1, output_padding=1, bias=True)
self.bn = nn.BatchNorm2d(noutput, eps=1e-3)
def forward(self, input):
output = self.conv(input)
output = self.bn(output)
return F.relu(output)
class PSPDec(nn.Module):
def __init__(self, in_features, out_features, downsize, upsize=(60,80)):
super(PSPDec,self).__init__()
self.features = nn.Sequential(
nn.AvgPool2d(downsize, stride=downsize),
nn.Conv2d(in_features, out_features, 1, bias=False),
nn.BatchNorm2d(out_features, momentum=.95),
nn.ReLU(inplace=True),
nn.Upsample(size=upsize, mode='bilinear')
)
def forward(self, x):
return self.features(x)
class Decoder (nn.Module):
def __init__(self, num_classes):
super().__init__()
self.layer5a = PSPDec(128, 32, (60,80),(60,80))
self.layer5b = PSPDec(128, 32, (30,40),(60,80))
self.layer5c = PSPDec(128, 32, (15,20),(60,80))
self.layer5d = PSPDec(128, 32, (int(7.5),10),(60,80))
self.final = nn.Sequential(
nn.Conv2d(256, 256, 3, padding=1, bias=False),
nn.BatchNorm2d(256, momentum=.95),
nn.ReLU(inplace=True),
nn.Dropout(.1),
nn.Conv2d(256, num_classes, 1),
)
def forward(self, x):
#x=x[0]
x = self.final(torch.cat([
x,
self.layer5a(x),
self.layer5b(x),
self.layer5c(x),
self.layer5d(x),
], 1))
#print('final', x.size())
return F.upsample(x,size=(480,640), mode='bilinear')
#ERFNet
class Net(nn.Module):
def __init__(self, num_classes, encoder=None): #use encoder to pass pretrained encoder
super().__init__()
if (encoder == None):
self.encoder = Encoder(num_classes)
else:
self.encoder = encoder
self.decoder = Decoder(num_classes)
def forward(self, input, only_encode=False):
if only_encode:
return self.encoder.forward(input, predict=True)
else:
output = self.encoder(input) #predict=False by default
return self.decoder.forward(output)