1only batches of spatial targets supported (non-empty 3D tensors) but got targets of size

因为在使用Crossentropyloss作为损失函数时，output=net(input)的output应该是[batchsize, channel, height, weight]，而label则是[batchsize, height, weight]，label是单通道灰度图，而在BCELoss中，两者都是[batchsize, channel, height, weight]。