Need help to solve a runtime error

[toseattle]

When I run the code, I keep getting the following error. Any suggestions would be appreciated!

Traceback (most recent call last):
File "/root/PycharmProjects/Photo-Sketching/train.py", line 68, in
main()
File "/root/PycharmProjects/Photo-Sketching/train.py", line 39, in main
model.optimize_parameters() # error
File "/root/PycharmProjects/Photo-Sketching/models/pix2pix_model.py", line 139, in optimize_parameters
self.backward_D()
File "/root/PycharmProjects/Photo-Sketching/models/pix2pix_model.py", line 106, in backward_D
pred_real = self.netD(real_AB) # error------------------------------------------------
File "/usr/local/lib/python3.5/dist-packages/torch/nn/modules/module.py", line 489, in call
result = self.forward(*input, **kwargs)
File "/root/PycharmProjects/Photo-Sketching/models/networks.py", line 423, in forward
return self.model(input)
File "/usr/local/lib/python3.5/dist-packages/torch/nn/modules/module.py", line 489, in call
result = self.forward(*input, **kwargs)
File "/usr/local/lib/python3.5/dist-packages/torch/nn/modules/container.py", line 92, in forward
input = module(input)
File "/usr/local/lib/python3.5/dist-packages/torch/nn/modules/module.py", line 489, in call
result = self.forward(*input, **kwargs)
File "/usr/local/lib/python3.5/dist-packages/torch/nn/modules/conv.py", line 320, in forward
self.padding, self.dilation, self.groups)
RuntimeError: Given groups=1, weight of size [64, 6, 4, 4], expected input[1, 4, 256, 256] to have 6 channels, but got 4 channels instead

[mtli]

If it is input and output related, it might be you have some errors in setting up these two options:
self.parser.add_argument('--input_nc', type=int, default=3, help='# of input image channels') self.parser.add_argument('--output_nc', type=int, default=3, help='# of output image channels')
Otherwise, you might want to check the shape of intermediate tensors, for example print(x.shape).

[toseattle]

Hi Mengtian,

Thanks much for your reply. I tried a couple of changes, but it still does not work. Specifically, the errors occur at: pred_fake = self.netD(fake_AB.detach()) (line 93 in your pix2pix_model.py in the models folder), and at: pred_real = self.netD(real_AB) (line 102 in pix2pix_model.py). The shape of 'fake_AB' is (1, 6, 256, 256) and the shape of 'real_AB' is (1, 4, 256, 256). I think the problem is no matter what discriminator we choose, the first conv layer needs to handle both 6 input channels and 4 input channels simultaneously, which might be impossible. Sorry to create more work for you, but I look forward to more suggestions or solutions. Thank you in advance.

[mtli]

Did you set --output_nc=1? The sketch output is not RGB, but grayscale (single channel).

[toseattle]

Change the output channel to 1 solves my problem!! You really helped me a lot, and thank you so much!!