-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathfinal_model_AGSSF1.py
278 lines (217 loc) · 10.7 KB
/
final_model_AGSSF1.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
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
from typing_extensions import Self
import torch
import torch.nn as nn
import torch.nn.functional as F
import numpy as np
import math
import cv2
import torch.fft as fft
def inv_mag(x):
fft_ = torch.fft.fft2(x)
fft_ = torch.fft.ifft2(1*torch.exp(1j*(fft_.angle())))
return fft_.real
class AGSSF(nn.Module):
def __init__(self, channels, b=1, gamma=2):
super(AGSSF, self).__init__()
self.avg_pool = nn.AdaptiveAvgPool2d(1)
self.channels = channels
self.b = b
self.gamma = gamma
self.conv = nn.Conv1d(1, 1, kernel_size=self.kernel_size(), padding=(self.kernel_size() - 1) // 2, bias=False)
self.sigmoid = nn.Sigmoid()
def kernel_size(self):
k = int(abs((math.log2(self.channels)/self.gamma)+ self.b/self.gamma))
out = k if k % 2 else k+1
return out
def forward(self, x):
x1=inv_mag(x)
# feature descriptor on the global spatial information
y = self.avg_pool(x1)
# Two different branches of ECA module
y = self.conv(y.squeeze(-1).transpose(-1, -2)).transpose(-1, -2).unsqueeze(-1)
# Multi-scale information fusion
y = self.sigmoid(y)
return x * y.expand_as(x)
class SFCA(nn.Module):
def __init__(self, channels, relu_slope=0.2, gamma=2):
super(SFCA, self).__init__()
self.identity1 = nn.Conv2d(channels, channels, 1)
self.identity2 = nn.Conv2d(channels, channels, 1)
self.conv_1 = nn.Conv2d(channels, 2*channels, kernel_size=1, bias=True)
self.relu_1 = nn.LeakyReLU(relu_slope)
self.conv_2 = nn.Conv2d(2*channels, channels, kernel_size=3, padding=1, groups=channels, bias=True)
self.relu_2 = nn.LeakyReLU(relu_slope)
self.conv_f1 = nn.Conv2d(channels, 2*channels, kernel_size=1)
self.conv_f2 = nn.Conv2d(2*channels, channels, kernel_size=1)
self.con2X1 = nn.Conv2d(2*channels, channels, kernel_size=1)
self.agssf = AGSSF(channels)
def forward(self, x):
out = self.conv_1(x)
out_1, out_2 = torch.chunk(out, 2, dim=1)
out = torch.cat([out_1, out_2], dim=1)
out = self.relu_1(out)
out = self.relu_2(self.conv_2(out))
# print(self.identity1(x).shape, out.shape)
out += self.identity1(x)
x_fft = fft.fftn(x, dim=(-2, -1)).real
x_fft = F.gelu(self.conv_f1(x_fft))
x_fft = self.conv_f2(x_fft)
x_reconstructed = fft.ifftn(x_fft, dim=(-2, -1)).real
x_reconstructed += self.identity2(x)
f_out = self.con2X1(torch.cat([out, x_reconstructed], dim=1))
return self.agssf(f_out)
class MDTA(nn.Module):
def __init__(self, channels, num_heads):
super(MDTA, self).__init__()
self.num_heads = num_heads
self.temperature = nn.Parameter(torch.ones(1, num_heads, 1, 1))
self.qkv = nn.Conv2d(channels, channels * 3, kernel_size=1, bias=False)
self.qkv_conv = nn.Conv2d(channels * 3, channels * 3, kernel_size=3, padding=1, groups=channels * 3, bias=False)
self.project_out = nn.Conv2d(channels, channels, kernel_size=1, bias=False)
#frequency
self.kv = nn.Conv2d(channels, channels * 2, kernel_size=1, bias=False)
self.q1X1_1 = nn.Conv2d(channels, channels , kernel_size=1, bias=False)
self.q1X1_2 = nn.Conv2d(channels, channels , kernel_size=1, bias=False)
self.kv_conv = nn.Conv2d(channels * 2, channels * 2, kernel_size=3, padding=1, groups=channels * 2, bias=False)
self.project_outf = nn.Conv2d(channels, channels, kernel_size=1, bias=False)
def forward(self, x):
b, c, h, w = x.shape
q, k, v = self.qkv_conv(self.qkv(x)).chunk(3, dim=1)
q = q.reshape(b, self.num_heads, -1, h * w)
k = k.reshape(b, self.num_heads, -1, h * w)
v = v.reshape(b, self.num_heads, -1, h * w)
q, k = F.normalize(q, dim=-1), F.normalize(k, dim=-1)
attn = torch.softmax(torch.matmul(q, k.transpose(-2, -1).contiguous()) * self.temperature, dim=-1)
out = self.project_out(torch.matmul(attn, v).reshape(b, -1, h, w))
# frequency
x_fft = fft.fftn(x, dim=(-2, -1)).real
x_fft=F.gelu(x_fft)
x_fft=self.q1X1_1(x_fft)
qf=fft.ifftn(x_fft,dim=(-2, -1)).real
kf, vf = self.kv_conv(self.kv(out)).chunk(2, dim=1)
qf = qf.reshape(b, self.num_heads, -1, h * w)
kf = kf.reshape(b, self.num_heads, -1, h * w)
vf = vf.reshape(b, self.num_heads, -1, h * w)
qf, kf = F.normalize(qf, dim=-1), F.normalize(kf, dim=-1)
attnf = torch.softmax(torch.matmul(qf, k.transpose(-2, -1).contiguous()) * self.temperature, dim=-1)
outf = self.project_outf(torch.matmul(attn, vf).reshape(b, -1, h, w))
return outf
class GDFN(nn.Module):
def __init__(self, channels, expansion_factor):
super(GDFN, self).__init__()
hidden_channels = int(channels * expansion_factor)
self.project_in = nn.Conv2d(channels, hidden_channels * 2, kernel_size=1, bias=False)
self.conv = nn.Conv2d(hidden_channels * 2, hidden_channels * 2, kernel_size=3, padding=1,
groups=hidden_channels * 2, bias=False)
self.project_out = nn.Conv2d(hidden_channels, channels, kernel_size=1, bias=False)
def forward(self, x):
x1, x2 = self.conv(self.project_in(x)).chunk(2, dim=1)
x = self.project_out(F.gelu(x1) * x2)
return x
class TransformerBlock(nn.Module):
def __init__(self, channels, num_heads, expansion_factor):
super(TransformerBlock, self).__init__()
self.norm1 = nn.LayerNorm(channels)
self.attn = MDTA(channels, num_heads)
self.norm2 = nn.LayerNorm(channels)
self.ffn = GDFN(channels, expansion_factor)
def forward(self, x):
b, c, h, w = x.shape
x = x + self.attn(self.norm1(x.reshape(b, c, -1).transpose(-2, -1).contiguous()).transpose(-2, -1)
.contiguous().reshape(b, c, h, w))
x = x + self.ffn(self.norm2(x.reshape(b, c, -1).transpose(-2, -1).contiguous()).transpose(-2, -1)
.contiguous().reshape(b, c, h, w))
return x
class DownSample(nn.Module):
def __init__(self, channels):
super(DownSample, self).__init__()
self.body = nn.Sequential(nn.Conv2d(channels, channels // 2, kernel_size=3, padding=1, bias=False),
nn.PixelUnshuffle(2))
def forward(self, x):
return self.body(x)
class UpSample(nn.Module):
def __init__(self, channels,channel_red):
super(UpSample, self).__init__()
self.amp_fuse = nn.Sequential(nn.Conv2d(channels,channels,1,1,0),nn.LeakyReLU(0.1,inplace=False),
nn.Conv2d(channels,channels,1,1,0))
self.pha_fuse = nn.Sequential(nn.Conv2d(channels,channels,1,1,0),nn.LeakyReLU(0.1,inplace=False),
nn.Conv2d(channels,channels,1,1,0))
if channel_red:
self.post = nn.Conv2d(channels, channels//2, 1, 1, 0)
else:
self.post = nn.Conv2d(channels, channels, 1, 1, 0)
def forward(self, x):
N, C, H, W = x.shape
fft_x = torch.fft.fft2(x)
mag_x = torch.abs(fft_x)
pha_x = torch.angle(fft_x)
Mag = self.amp_fuse(mag_x)
Pha = self.pha_fuse(pha_x)
amp_fuse = torch.tile(Mag, (2, 2))
pha_fuse = torch.tile(Pha, (2, 2))
real = amp_fuse * torch.cos(pha_fuse)
imag = amp_fuse * torch.sin(pha_fuse)
out = torch.complex(real, imag)
output = torch.fft.ifft2(out)
output = torch.abs(output)
return self.post(output)
class UpSample1(nn.Module):
def __init__(self, channels):
super(UpSample1, self).__init__()
self.body = nn.Sequential(nn.Conv2d(channels, channels * 2, kernel_size=3, padding=1, bias=False),
nn.PixelShuffle(2))
def forward(self, x):
return self.body(x)
class UpS(nn.Module):
def __init__(self, channels):
super(UpS, self).__init__()
self.Fups=UpSample(channels,True)
self.Sups=UpSample1(channels)
self.reduce=nn.Conv2d(channels, channels // 2, kernel_size=1,bias=False)
def forward(self, x):
out=torch.cat([self.Fups(x),self.Sups(x)],dim=1)
# print(out.shape)
return self.reduce(out)
class mymodel(nn.Module):
def __init__(self, num_blocks=[2, 3, 3, 4], num_heads=[1, 2, 4, 8], channels=[16, 32, 64, 128], num_refinement=4,
expansion_factor=2.66, ch=[64,32,16,64]):
super(mymodel, self).__init__()
self.attention = nn.ModuleList([SFCA(num_ch) for num_ch in ch])
self.embed_conv_rgb = nn.Conv2d(3, channels[0], kernel_size=3, padding=1, bias=False)
self.encoders = nn.ModuleList([nn.Sequential(*[TransformerBlock(num_ch, num_ah, expansion_factor) for _ in range(num_tb)]) for num_tb, num_ah, num_ch in
zip(num_blocks, num_heads, channels)])
self.down1 = DownSample(channels[0])
self.down2 = DownSample(channels[1])
self.down3 = DownSample(channels[2])
self.ups_1=UpS(128)
self.ups_2=UpS(64)
self.ups_3=UpS(32)
self.ups_4=UpS(3)
self.ups1 = UpSample1(32)
self.reduces2 = nn.Conv2d(64, 32, kernel_size=1, bias=False)
self.reduces1=nn.Conv2d(128, 64, kernel_size=1, bias=False)
self.decoders = nn.ModuleList([nn.Sequential(*[TransformerBlock(channels[2], num_heads[2], expansion_factor)
for _ in range(num_blocks[2])])])
self.decoders.append(nn.Sequential(*[TransformerBlock(channels[1], num_heads[1], expansion_factor)
for _ in range(num_blocks[1])]))
self.decoders.append(nn.Sequential(*[TransformerBlock(channels[1], num_heads[0], expansion_factor) for _ in range(num_blocks[0])]))
self.refinement = nn.Sequential(*[TransformerBlock(channels[1], num_heads[0], expansion_factor)
for _ in range(num_refinement)])
self.output = nn.Conv2d(8, 3, kernel_size=3, padding=1, bias=False)
self.output1= nn.Conv2d(16, 8, kernel_size=3, padding=1, bias=False)
self.ups2 = UpSample1(16)
self.outputl=nn.Conv2d(32, 8, kernel_size=3, padding=1, bias=False)
def forward(self,RGB_input):
###-------encoder for RGB-------####
fo_rgb = self.embed_conv_rgb(RGB_input)
out_enc_rgb1 = self.encoders[0](fo_rgb)
out_enc_rgb2 = self.encoders[1](self.down1(out_enc_rgb1))
out_enc_rgb3 = self.encoders[2](self.down2(out_enc_rgb2))
out_enc_rgb4 = self.encoders[3](self.down3(out_enc_rgb3))
###-------Dencoder------###
out_dec3 = self.decoders[0](self.reduces1(torch.cat([(self.ups_1(out_enc_rgb4)), self.attention[0](out_enc_rgb3)], dim=1)))
out_dec2 = self.decoders[1](self.reduces2(torch.cat([self.ups_2(out_dec3),self.attention[1](out_enc_rgb2)], dim=1)))
fd = self.decoders[2](torch.cat([self.ups_3(out_dec2),self.attention[2](out_enc_rgb1)], dim=1))
fr = self.refinement(fd)
outi=self.ups1(fr)
return self.output(self.outputl(fr))