Two bugs encountered

[kaufManu]

Great work - thanks a lot for this!

When using the code on some test images, I had to adjust the following pieces of code:

In Net.forward() after line 135 I added the statement

is_max = torch.unsqueeze(is_max, dim=0)

In Net.__init__() change line 79 to prevent a dimension mismatch as follows:

self.directional_filter.weight.data.copy_(torch.from_numpy(all_filters[:, None, ...]))

[DCurro]

I've taken your changes into consideration. This code was written in a very old version of PyTorch. I've updated to the newest version, and provided a pip requirements.txt file.

Thank you for letting me know.