def __init__(self, version=0):
super(VGGBase, self).__init__()
self.conv1 = nn.Conv2d(56, 512, kernel_size=3, padding=1, bias=True)
self.conv1_bn = nn.BatchNorm2d(512, affine=True)
self.conv2 = nn.Conv2d(112, 512, kernel_size=3, padding=1, bias=True)
self.conv2_bn = nn.BatchNorm2d(512, affine=True)
self.conv3 = nn.Conv2d(160, 512, kernel_size=3, padding=1, bias=True)
self.conv3_bn = nn.BatchNorm2d(512, affine=True)
self.deconv6 = nn.Conv2d(448, 272, kernel_size=3, padding=1, bias=True)
self.deconv6_bn = nn.BatchNorm2d(272, affine=True)
self.deconv5 = nn.Conv2d(272, 160, kernel_size=3, padding=1, bias=True)
self.deconv5_bn = nn.BatchNorm2d(160, affine=True)
self.deconv4 = nn.Conv2d(160, 112, kernel_size=3, padding=1, bias=True)
self.deconv4_bn = nn.BatchNorm2d(112, affine=True)
self.deconv3 = nn.Conv2d(112, 56, kernel_size=3, padding=1, bias=True)
self.deconv3_bn = nn.BatchNorm2d(56, affine=True)
self.efficientNet = EfficientNet.from_pretrained('efficientnet-b4')
state_dict = self.efficientNet.state_dict()
self.load_state_dict(state_dict, strict=False)
def __init__(self, version=0):
super(VGGBase, self).__init__()
self.efficientNet = EfficientNet.from_pretrained('efficientnet-b4')
state_dict = self.efficientNet.state_dict()
self.load_state_dict(state_dict, strict=False)
def __init__(self, version=0):
super(VGGBase, self).__init__()
self.w1 = torch.nn.Parameter(torch.ones(1))
self.w1_2 = torch.nn.Parameter(torch.ones(1))
self.eps = 0.0001
self.w2 = torch.nn.Parameter(torch.ones(1))
self.w2_2 = torch.nn.Parameter(torch.ones(1))
self.w3 = torch.nn.Parameter(torch.ones(1))
self.w3_2 = torch.nn.Parameter(torch.ones(1))
self.w4 = torch.nn.Parameter(torch.ones(1))
self.w4_2 = torch.nn.Parameter(torch.ones(1))
self.conv1 = nn.Conv2d(56, 512, kernel_size=3, padding=1, bias=True)
self.conv1_bn = nn.BatchNorm2d(512, affine=True)
self.conv2 = nn.Conv2d(112, 512, kernel_size=3, padding=1, bias=True)
self.conv2_bn = nn.BatchNorm2d(512, affine=True)
self.conv3 = nn.Conv2d(160, 512, kernel_size=3, padding=1, bias=True)
self.conv3_bn = nn.BatchNorm2d(512, affine=True)
# self.deconv6=nn.Conv2d(448,272,kernel_size=3,padding=1,bias=True)
# self.deconv6_bn= nn.BatchNorm2d(272, affine=True)
# self.deconv5=nn.Conv2d(272,160,kernel_size=3,padding=1,bias=True)
# self.deconv5_bn= nn.BatchNorm2d(160, affine=True)
# self.deconv4=nn.Conv2d(160,112,kernel_size=3,padding=1,bias=True)
# self.deconv4_bn= nn.BatchNorm2d(112, affine=True)
# self.deconv3=nn.Conv2d(112,56,kernel_size=3,padding=1,bias=True)
# self.deconv3_bn= nn.BatchNorm2d(56, affine=True)
self.deconv6 = nn.ConvTranspose2d(448, 272, kernel_size=2, stride=2)
self.deconv6_bn = nn.BatchNorm2d(272, affine=True)
self.deconv5 = nn.ConvTranspose2d(272, 160, kernel_size=2, stride=2)
self.deconv5_bn = nn.BatchNorm2d(160, affine=True)
self.deconv4 = nn.ConvTranspose2d(160, 112, kernel_size=2, stride=2)
self.deconv4_bn = nn.BatchNorm2d(112, affine=True)
self.deconv3 = nn.ConvTranspose2d(112, 56, kernel_size=2, stride=2)
self.deconv3_bn = nn.BatchNorm2d(56, affine=True)
self.efficientNet = EfficientNet.from_pretrained('efficientnet-b4')
state_dict = self.efficientNet.state_dict()
self.load_state_dict(state_dict, strict=False)