def __init__(self):
super(DSSNet, self).__init__()
self.resnet = torchvision.models.resnet34(True)
self.conv1 = nn.Sequential(
self.resnet.conv1,
self.resnet.bn1,
self.resnet.relu)
self.encode2 = nn.Sequential(self.resnet.layer1,
SCSE(64))
self.encode3 = nn.Sequential(self.resnet.layer2,
SCSE(128))
self.encode4 = nn.Sequential(self.resnet.layer3,
SCSE(256))
self.encode5 = nn.Sequential(self.resnet.layer4,
SCSE(512))
self.center = nn.Sequential(SeRes(512),
SeRes(512),
BasicConv2d(512, 256, 3, 1),
nn.MaxPool2d(2, 2))
self.decode5 = Decoderv2(256, 512, 64)
self.decode4 = Decoderv2(64, 256, 64)
self.decode3 = Decoderv2(64, 128, 64)
self.decode2 = Decoderv2(64, 64, 64)
self.decode1 = Decoder(64, 64)
self.logit = nn.Sequential(nn.Conv2d(4, 1, kernel_size=1, bias=False))
self.logit1 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1, bias=False))
self.logit2 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1, bias=False))
self.logit3 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1, bias=False))
self.logit4 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1, bias=False))
def __init__(self):
super(AUNet, self).__init__()
self.xception = xception()
self.encode1 = nn.Sequential(self.xception.conv1, self.xception.bn1,
self.xception.relu1, self.xception.conv2,
self.xception.bn2, self.xception.relu2)
self.encode2 = self.xception.block1
self.encode3 = self.xception.block2
self.encode4 = self.xception.block3
self.encode5 = nn.Sequential(
self.xception.block4, self.xception.block5, self.xception.block6,
self.xception.block7, self.xception.block8, self.xception.block9,
self.xception.block10, self.xception.block11)
self.center = nn.Sequential(SeRes(728), SeRes(728),
BasicConv2d(728, 256, 3, 1),
nn.MaxPool2d(2, 2))
self.decode5 = Decoderv2(256, 728, 64)
self.decode4 = Decoderv2(64, 256, 64)
self.decode3 = Decoderv2(64, 128, 64)
self.decode2 = Decoderv2(64, 64, 64)
self.decode1 = Decoder(64, 64)
self.logit = nn.Sequential(nn.Conv2d(4, 1, kernel_size=1, bias=False))
self.logit1 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1,
bias=False))
self.logit2 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1,
bias=False))
self.logit3 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1,
bias=False))
self.logit4 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1,
bias=False))
def __init__(self, x1_c, x2_c, out_c, pad = None):
super(Decoderv2, self).__init__()
self.ct_out = x1_c // 2
self.x2_c = x2_c
self.ct_conv = nn.ConvTranspose2d(x1_c, self.ct_out, 2, stride=2)
self.conx2 = BasicConv2d(x2_c, self.ct_out, 3, 1)
self.res1 = SeRes(x1_c)
self.res2 = SeRes(x1_c)
self.con = BasicConv2d(x1_c, out_c, 3, 1)
def __init__(self):
super(AUNet, self).__init__()
self.vgg = torchvision.models.vgg19(True)
#128
self.encode2 = nn.Sequential(self.vgg.features[0],
self.vgg.features[1],
self.vgg.features[2],
self.vgg.features[3],
self.vgg.features[4])
#64
self.encode3 = nn.Sequential(self.vgg.features[5],
self.vgg.features[6],
self.vgg.features[7],
self.vgg.features[8],
self.vgg.features[9])
#32
self.encode4 = nn.Sequential(
self.vgg.features[10], self.vgg.features[11],
self.vgg.features[12], self.vgg.features[13],
self.vgg.features[14], self.vgg.features[15],
self.vgg.features[16], self.vgg.features[17],
self.vgg.features[18])
#16
self.encode5 = nn.Sequential(
self.vgg.features[19], self.vgg.features[20],
self.vgg.features[21], self.vgg.features[22],
self.vgg.features[23], self.vgg.features[24],
self.vgg.features[25], self.vgg.features[26],
self.vgg.features[27])
self.center = nn.Sequential(SeRes(512), SeRes(512),
BasicConv2d(512, 256, 3, 1),
nn.MaxPool2d(2, 2))
#
self.decode5 = Decoderv2(256, 512, 64)
self.decode4 = Decoderv2(64, 256, 64)
self.decode3 = Decoderv2(64, 128, 64)
self.decode2 = Decoderv2(64, 64, 64)
self.decode1 = Decoder(64, 64)
self.logit = nn.Sequential(nn.Conv2d(4, 1, kernel_size=1, bias=False))
self.logit1 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1,
bias=False))
self.logit2 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1,
bias=False))
self.logit3 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1,
bias=False))
self.logit4 = nn.Sequential(nn.Conv2d(64, 1, kernel_size=1,
bias=False))
def __init__(self, x1_c, out_c):
super(Decoder, self).__init__()
self.ct_conv = nn.ConvTranspose2d(x1_c, out_c, 2, stride=2)
self.res1 = SeRes(out_c, 8)
self.res2 = SeRes(out_c, 8)