Run time error.output with shape [1, 256, 192] doesn't match the broadcast shape [3, 256, 192]

Question

Run time error.output with shape [1, 256, 192] doesn't match the broadcast shape [3, 256, 192]

saisusrith opened this issue 4 years ago · comments

Hi while running the your template.ipynb we get the following error.
Even after debugging I could not find the error.
RuntimeError: output with shape [1, 256, 192] doesn't match the broadcast shape [3, 256, 192]

Guanglei Xu · Answer 1 · Sun Jan 26 2020 22:48:06 GMT+0800 (China Standard Time)

Hi, thanks for this issue.

Can you copy complete error logs, like which line incurred this error?

saisusrith · Answer 2 · Mon Jan 27 2020 08:40:10 GMT+0800 (China Standard Time)

Hi GrayXu,we debugged your code . The error is at line 179 "one_map = Model.transformer(one_map)".After it calls the Model.transformer error pops up

saisusrith · Answer 3 · Thu Jan 30 2020 02:59:20 GMT+0800 (China Standard Time)

Hi @GrayXu any solution?

Guanglei Xu · Answer 4 · Fri Feb 21 2020 22:52:02 GMT+0800 (China Standard Time)

Sry,
because of Coronavirus vacation in China, I can't access GPU machines to test codes these days.
I don't find errors in Model.transformer. The transformer' just tensorlize and normalize data.
Can you show your local enviroments? I only test it on torch==1.2.0 torchvision==0.2.0 with GPU.

carfei · Answer 5 · Thu Apr 16 2020 13:37:51 GMT+0800 (China Standard Time)

I also encountered the same error message, don't know if you solved it? How to solve it?

carfei · Answer 6 · Thu Apr 16 2020 14:52:37 GMT+0800 (China Standard Time)

@GrayXu ，An error occurred in getPoseMap(self, pose_data)，The number of output channels changed from 3 to 1.So how to solve it?

mad-fogs · Answer 7 · Thu Jul 02 2020 19:43:55 GMT+0800 (China Standard Time)

Sry,
because of Coronavirus vacation in China, I can't access GPU machines to test codes these days.
I don't find errors in Model.transformer. The transformer' just tensorlize and normalize data.
Can you show your local enviroments? I only test it on torch==1.2.0 torchvision==0.2.0 with GPU.

i use torch==1.2.0 torchvision==0.2.0 with GPU, but the error still exists.

mad-fogs · Answer 8 · Thu Jul 02 2020 22:01:36 GMT+0800 (China Standard Time)

Sry,
because of Coronavirus vacation in China, I can't access GPU machines to test codes these days.
I don't find errors in Model.transformer. The transformer' just tensorlize and normalize data.
Can you show your local enviroments? I only test it on torch==1.2.0 torchvision==0.2.0 with GPU.

i use torch==1.2.0 torchvision==0.2.0 with GPU, but the error still exists.

solved. input image size should be 192*256.

Ananta Arora · Answer 9 · Sun Nov 08 2020 03:07:52 GMT+0800 (China Standard Time)

Hi @mad-fogs ,
When I am clicking the button for selecting the cloth, it throws an error on submitting. We are saving them in static/img folder right?