PSMNet debug(二)

submodule.py

1.

1
3
384
1248

def convbn(in_planes, out_planes, kernel_size, stride, pad, dilation):

    return nn.Sequential(nn.Conv2d(in_planes, out_planes, kernel_size=kernel_size, stride=stride, padding=dilation if dilation > 1 else pad, dilation = dilation, bias=False),
                         nn.BatchNorm2d(out_planes))
 

class feature_extraction(nn.Module):

    def __init__(self):
        print('feature_extraction 123')
        super(feature_extraction, self).__init__()
        self.inplanes = 32
        self.firstconv = nn.Sequential(convbn(3, 32, 3, 2, 1, 1), #下采样，缩小一倍。w,h各除以2.filter num:32。
                                       nn.ReLU(inplace=True),
                                       convbn(32, 32, 3, 1, 1, 1),
                                       nn.ReLU(inplace=True),
                                       convbn(32, 32, 3, 1, 1, 1),
                                       nn.ReLU(inplace=True))

对应下面。

def forward(self, x):

        output      = self.firstconv(x)
        print('submodule firstconv:')
        print(output.size()[0])
        print(output.size()[1])
        print(output.size()[2])
        print(output.size()[3])

submodule firstconv:
1
32
192
624

 

2.

        self.lastconv = nn.Sequential(convbn(320, 128, 3, 1, 1, 1),
                                      nn.ReLU(inplace=True),
                                      nn.Conv2d(128, 32, kernel_size=1, padding=0, stride = 1, bias=False))

        output_feature = torch.cat((output_raw, output_skip, output_branch4, output_branch3, output_branch2, output_branch1),

#output_branch4, output_branch3, output_branch2, output_branch1 第2个通道分别是32，output_raw是64，output_skip是128。其