In train_RINDNet.py, size mismatch error

[carwyn987]

Line 89:

loss = self.criterion([out_depth,out_normal,out_reflectance,out_illumination], target[:, 1:5, :, :])

I get the following error:

ValueError: Using a target size (torch.Size([4, 320, 320])) that is different to the input size (torch.Size([5, 1, 320, 320])) is deprecated. Please ensure they have the same size.

Fixed by replacing line with:

loss = self.criterion(torch.cat((out_depth,out_normal,out_reflectance,out_illumination),axis=1), target[:, 1:5, :, :])

Unsure if this fix breaks training, but fixes error and allows training to commence.

Using Python 3.7, numpy 1.19.5, torch 1.13.1, opencv-python 4.3.0.38, ...

[MengyangPu]

Thanks, this fix will not affect training, or replacing line with:
loss = self.criterion(torch.cat((out_depth,out_normal,out_reflectance,out_illumination),dim=1), target[:, 1:5, :, :])