RuntimeError：给定 groups=1，权重大小为 [32, 3, 16, 16, 16]，预期输入 [100, 16, 16, 16, 3] 有 3 个通道，但得到了 16 个通道

Question

RuntimeError：给定 groups=1，权重大小为 [32, 3, 16, 16, 16]，预期输入 [100, 16, 16, 16, 3] 有 3 个通道，但得到了 16 个通道

这是我认为问题所在的代码部分。

def __init__(self):
        super(Lightning_CNNModel, self).__init__()

        self.conv_layer1 = self._conv_layer_set(3, 32)
        self.conv_layer2 = self._conv_layer_set(32, 64)
        self.fc1 = nn.Linear(2**3*64, 128)
        self.fc2 = nn.Linear(128, 10)   # num_classes = 10
        self.relu = nn.LeakyReLU()
        self.batch=nn.BatchNorm1d(128)
        self.drop=nn.Dropout(p=0.15)

    def _conv_layer_set(self, in_c, out_c):
        conv_layer = nn.Sequential(
            nn.Conv3d(in_c, out_c, kernel_size=(3, 3, 3), padding=0),
            nn.LeakyReLU(),
            nn.MaxPool3d((2, 2, 2)),
        )

        return conv_layer



    def forward(self, x):
        out = self.conv_layer1(x)
        out = self.conv_layer2(out)
        out = out.view(out.size(0), -1)
        out = self.fc1(out)
        out = self.relu(out)
        out = self.batch(out)
        out = self.drop(out)
        out = self.fc2(out)

        return out

这是我正在处理的代码

Answer 1

nn.Conv3d期望输入的大小为[batch_size,channels,deep,height,width]。第一个卷积需要 3 个通道，但如果您的输入大小为[100, 16 , 16, 16, 3]，则为 16 个通道。

假设您的数据给出为[batch_size, Depth, Height, width, Channels]，您需要交换尺寸，这可以通过以下方式完成torch.Tensor.permute：

# From: [batch_size, depth, height, width, channels]
# To: [batch_size, channels, depth, height, width]
input = input.permute(0, 4, 1, 2, 3)