|
|
from rscd.models.decoderheads.lgpnet.unet_parts import *
|
|
|
from rscd.models.decoderheads.lgpnet.ChannelAttention import ChannelAttention
|
|
|
from rscd.models.decoderheads.lgpnet.SpatialPyramidModule import SPM
|
|
|
from rscd.models.decoderheads.lgpnet.FeaturePyramidModule import FPM
|
|
|
from rscd.models.decoderheads.lgpnet.PositionAttentionModule import PAM
|
|
|
|
|
|
class BFExtractor(nn.Module):
|
|
|
""" Full assembly of the parts to form the complete network """
|
|
|
def __init__(self, n_channels, n_classes, bilinear=False):
|
|
|
super(BFExtractor, self).__init__()
|
|
|
self.n_channels = n_channels
|
|
|
self.n_classes = n_classes
|
|
|
self.bilinear = bilinear
|
|
|
|
|
|
self.inc = DoubleConv(n_channels, 64)
|
|
|
self.down1 = Down(64, 128)
|
|
|
self.down2 = Down(128, 256)
|
|
|
self.down3 = Down(256, 512)
|
|
|
self.down4 = Down(512, 1024)
|
|
|
|
|
|
self.pam = PAM(1024)
|
|
|
|
|
|
self.psp = SPM(1024, 1024, sizes=(1, 2, 3, 6))
|
|
|
self.fpa = FPM(1024)
|
|
|
self.drop = nn.Dropout2d(p=0.2)
|
|
|
self.ca = ChannelAttention(in_channels=1024)
|
|
|
self.conv1x1 = nn.Conv2d(1024, 512, kernel_size=1, stride=1, bias=False)
|
|
|
|
|
|
self.up1 = Up(1024, 512, bilinear)
|
|
|
self.ca1 = ChannelAttention(in_channels=512)
|
|
|
self.up2 = Up(512, 256, bilinear)
|
|
|
self.ca2 = ChannelAttention(in_channels=256)
|
|
|
self.up3 = Up(256, 128, bilinear)
|
|
|
self.ca3 = ChannelAttention(in_channels=128)
|
|
|
self.up4 = Up(128, 64, bilinear)
|
|
|
self.ca4 = ChannelAttention(in_channels=64)
|
|
|
self.outc = OutConv(64, n_classes)
|
|
|
|
|
|
def forward(self, x):
|
|
|
x1 = self.inc(x)
|
|
|
x2 = self.down1(x1)
|
|
|
x3 = self.down2(x2)
|
|
|
x4 = self.down3(x3)
|
|
|
x5 = self.down4(x4)
|
|
|
pam_x5 = self.pam(x5)
|
|
|
|
|
|
|
|
|
psp = self.psp(pam_x5)
|
|
|
pspdrop = self.drop(psp)
|
|
|
capsp = self.ca(pspdrop)
|
|
|
capsp = self.conv1x1(capsp)
|
|
|
|
|
|
|
|
|
fpa = self.fpa(pam_x5)
|
|
|
fpadrop = self.drop(fpa)
|
|
|
cafpa = self.ca(fpadrop)
|
|
|
cafpa = self.conv1x1(cafpa)
|
|
|
|
|
|
ca_psp_fpa = torch.cat([capsp, cafpa], dim=1)
|
|
|
|
|
|
x = self.up1(ca_psp_fpa, x4)
|
|
|
x = self.ca1(x)
|
|
|
x = self.up2(x, x3)
|
|
|
x = self.ca2(x)
|
|
|
x = self.up3(x, x2)
|
|
|
x = self.ca3(x)
|
|
|
x = self.up4(x, x1)
|
|
|
feats = self.ca4(x)
|
|
|
logits = self.outc(x)
|
|
|
return logits, feats
|
|
|
|
|
|
if __name__ == '__main__':
|
|
|
net = BFExtractor(n_channels=3, n_classes=1)
|
|
|
print(net) |
