stack expects each tensor

[sanazsab]

File ".../utils/util.py", line 100, in depth2normal_ortho
p = torch.stack([x_cord, y_cord, depth], dim=3) # (B,H,W,3)
RuntimeError: stack expects each tensor to be equal size, but got [8, 256, 192] at entry 0 and [5, 256, 192] at entry 2

I've got this error when I used batch size 8 for DRM.

What is the batch size for DRM?

Thanks for your quick answer.

[fyviezhao]

It is a bit weird. Given batch = 8 for DRM, the output depth should have 8 channels instead of 5. Have you modified other parts of the code that may lead to this error?