增大感受野

 

 self.stage1 = nn.Sequential(
            conv_bn(3, 64, 2, leaky = 0.1),    # 3
            conv_dw(64, 96, 1),   # 7
            conv_dw(96, 96, 2),  # 11
            conv_dw(96, 128, 1),  # 19
            conv_dw(128, 128, 2),  # 27
            conv_dw(128, 144, 1),  # 43
        )

# import mish
def conv_bn(inp, oup, stride = 1, leaky = 0):
    return nn.Sequential(
        nn.Conv2d(inp, oup, 3, stride, 1, bias=False),
        nn.BatchNorm2d(oup),
        # mish.Mish()
        nn.ReLU6()
    )

 

1.

x = F.max_pool2d(x, kernel_size=3, stride=2, padding=1)

self.conv2 = nn.Conv2d(96, 96, kernel_size=3, stride=2, padding=1)

self.conv4_0 = nn.Conv2d(128, 128, kernel_size=5, stride=2, padding=2)

 

增大感受野的方法
主要的方法是从增加网络的深度出发（这也