-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathoperations.py
142 lines (113 loc) · 5.69 KB
/
operations.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
import torch
import torch.nn as nn
from torch.autograd import Variable
from collections import OrderedDict
OPS = {
'none' : lambda C, stride, affine, edge: Zero(stride),
'noise': lambda C, stride, affine, edge: NoiseOp(stride, 0., 1.),
'skip_connect' : lambda C, stride, affine, edge: Identity() if stride == 1 else FactorizedReduce(C, C, edge, affine=affine),
'max_pool_3x3' : lambda C, stride, affine, edge: layers.MaxPooling2D((3, 3), strides=stride, padding='SAME'),
'avg_pool_3x3' : lambda C, stride, affine, edge: layers.AvgPooling2D((3, 3), strides=stride, padding='SAME'),
'sep_conv_3x3' : lambda C, stride, affine, edge: SepConv(C, C, 3, stride, 1, edge, affine=affine),
'sep_conv_5x5' : lambda C, stride, affine, edge: SepConv(C, C, 5, stride, 2, edge, affine=affine),
'sep_conv_7x7' : lambda C, stride, affine, edge: SepConv(C, C, 7, stride, 3, edge, affine=affine),
'dil_conv_3x3' : lambda C, stride, affine, edge: DilConv(C, C, 3, stride, 2, 2, edge, affine=affine),
'dil_conv_5x5' : lambda C, stride, affine, edge: DilConv(C, C, 5, stride, 4, 2, edge, affine=affine),
'conv_7x1_1x7' : lambda C, stride, affine, edge: nn.Sequential(
nn.ReLU(inplace=False),
nn.Conv2d(C, C, (1,7), stride=(1, stride), padding=(0, 3), bias=False),
nn.Conv2d(C, C, (7,1), stride=(stride, 1), padding=(3, 0), bias=False),
nn.BatchNorm2d(C, affine=affine)
),
}
class NoiseOp(nn.Module):
def __init__(self, stride, mean, std):
super(NoiseOp, self).__init__()
self.stride = stride
self.mean = mean
self.std = std
def forward(self, x):
if self.stride != 1:
x_new = x[:,:,::self.stride,::self.stride]
else:
x_new = x
noise = Variable(x_new.data.new(x_new.size()).normal_(self.mean, self.std))
return noise
class ReLUConvBN(nn.Module):
def __init__(self, C_in, C_out, kernel_size, stride, padding, edge, affine=True):
super(ReLUConvBN, self).__init__()
self.op = nn.Sequential(OrderedDict([('relu',
nn.ReLU(inplace=False)),('reluconvbnk'+str(kernel_size)+'_convkxk_edge-'+str(edge),
nn.Conv2d(C_in, C_out, kernel_size, stride=stride, padding=padding, bias=False)), ('reluconvbnk'+str(kernel_size)+'_bn_edge-'+str(edge),
nn.BatchNorm2d(C_out, affine=affine))])
)
def forward(self, x):
return self.op(x)
class DilConv(nn.Module):
def __init__(self, C_in, C_out, kernel_size, stride, padding, dilation, edge, affine=True):
super(DilConv, self).__init__()
self.op = nn.Sequential(OrderedDict([('relu',
nn.ReLU(inplace=False)), ('dilconvk'+str(kernel_size)+'_convkxk_edge-'+str(edge),
nn.Conv2d(C_in, C_in, kernel_size=kernel_size, stride=stride, padding=padding, dilation=dilation, groups=C_in, bias=False)), ('dilconvk'+str(kernel_size)+'_conv1x1_edge-'+str(edge),
nn.Conv2d(C_in, C_out, kernel_size=1, padding=0, bias=False)), ('dilconvk'+str(kernel_size)+'_bn_edge-'+str(edge),
nn.BatchNorm2d(C_out, affine=affine))])
)
def forward(self, x):
return self.op(x)
class SepConv(nn.Module):
def __init__(self, C_in, C_out, kernel_size, stride, padding, edge, affine=True):
super(SepConv, self).__init__()
self.op = nn.Sequential(OrderedDict([('relu0',
nn.ReLU(inplace=False)),('sepconvk'+str(kernel_size) + '_convkxk-0_edge-' + str(edge),
nn.Conv2d(C_in, C_in, kernel_size=kernel_size, stride=stride, padding=padding, groups=C_in, bias=False)), ('sepconvk'+str(kernel_size) + '_conv1x1-0_edge-' +str(edge),
nn.Conv2d(C_in, C_in, kernel_size=1, padding=0, bias=False)), ('sepconvk'+str(kernel_size) + '_bn-0_edge-' +str(edge),
nn.BatchNorm2d(C_in, affine=affine)), ('relu1',
nn.ReLU(inplace=False)), ('sepconvk'+str(kernel_size) + '_convkxk-1_edge-' +str(edge),
nn.Conv2d(C_in, C_in, kernel_size=kernel_size, stride=1, padding=padding, groups=C_in, bias=False)), ('sepconvk'+str(kernel_size) + '_conv1x1-1_edge-' +str(edge),
nn.Conv2d(C_in, C_out, kernel_size=1, padding=0, bias=False)), ('sepconvk'+str(kernel_size) + '_bn-1_edge-' +str(edge),
nn.BatchNorm2d(C_out, affine=affine))])
)
def forward(self, x):
return self.op(x)
class Identity(nn.Module):
def __init__(self):
super(Identity, self).__init__()
def forward(self, x):
return x
'''
class Zero(nn.Module):
def __init__(self, stride):
super(Zero, self).__init__()
self.stride = stride
def forward(self, x):
if self.stride == 1:
return x.mul(0.)
return x[:,:,::self.stride,::self.stride].mul(0.)
'''
class Zero(nn.Module):
def __init__(self, stride):
super(Zero, self).__init__()
self.stride = stride
def forward(self, x):
n, c, h, w = x.size()
h //= self.stride
w //= self.stride
if x.is_cuda:
with torch.cuda.device(x.get_device()):
padding = torch.cuda.FloatTensor(n, c, h, w).fill_(0)
else:
padding = torch.FloatTensor(n, c, h, w).fill_(0)
return padding
class FactorizedReduce(nn.Module):
def __init__(self, C_in, C_out, edge, affine=True):
super(FactorizedReduce, self).__init__()
assert C_out % 2 == 0
self.relu = nn.Sequential(OrderedDict([('skip_relu_' + str(edge), nn.ReLU(inplace=False))]))
self.conv_1 = nn.Sequential(OrderedDict([('skip_conv1_'+str(edge), nn.Conv2d(C_in, C_out // 2, 1, stride=2, padding=0, bias=False))]))
self.conv_2 = nn.Sequential(OrderedDict([('skip_conv2_'+str(edge), nn.Conv2d(C_in, C_out // 2, 1, stride=2, padding=0, bias=False))]))
self.bn = nn.Sequential(OrderedDict([('skip_bn_'+str(edge), nn.BatchNorm2d(C_out, affine=affine))]))
def forward(self, x):
x = self.relu(x)
out = torch.cat([self.conv_1(x), self.conv_2(x[:,:,1:,1:])], dim=1)
out = self.bn(out)
return out