search

minar09 / ACGPN

"Towards Photo-Realistic Virtual Try-On by Adaptively Generating↔Preserving Image Content"，CVPR 2020. (Modified from original with fixes for inference)
https://github.com/switchablenorms/DeepFashion_Try_On
77 stars 70 forks source link

Label error #4

Open josearangos opened 4 years ago

josearangos commented 4 years ago

hello again, thank you very much for the previous answer.

I am trying to run the test.py file with custom images and I am getting the following error:

File "test.py", line 126, in main() File "test.py", line 103, in main mask_clothes.cuda()), Variable(data['color'].cuda()), Variable(all_clothes_label.cuda()), Variable(data['image'].cuda()), Variable(data['pose'].cuda()), Variable(data['image'].cuda()), Variable(mask_fore.cuda())) File "/home/josearangos/anaconda3/envs/torch_env/lib/python3.6/site-packages/torch/nn/modules/module.py", line 477, in call result = self.forward(*input, **kwargs) File "/home/josearangos/Documentos/Clothe_mE/repositorios/versions_DeepFashion_Try_On/ACGPN/models/pix2pixHD_model.py", line 493, in forward return self.inference(label, pre_clothes_mask, img_fore, clothes_mask, clothes, all_clothes_label, real_image, pose, grid, mask_fore) File "/home/josearangos/Documentos/Clothe_mE/repositorios/versions_DeepFashion_Try_On/ACGPN/models/pix2pixHD_model.py", line 388, in inference (label.cpu().numpy() == 11).astype(np.float)).cuda() RuntimeError: cuda runtime error (59) : device-side assert triggered at /pytorch/aten/src/THC/generic/THCTensorCopy.cpp:70

Thank you for helping me

minar09 commented 4 years ago

Hi @josearangos , This is probably because you segmentation label has undefined values. See here: https://github.com/pytorch/pytorch/issues/1204#issuecomment-292746566 Please check your files and debug to see where its getting such values.

minar09 commented 4 years ago

Or sometimes this error happens for labels mismatch.

josearangos commented 4 years ago

@minar09 Thanks for the recommendation, thanks to you I can run the inference.

These are the results:

Why would you put those colors on the nose?

image image image image

what do you think of this result?

image

minar09 commented 4 years ago

Hi @josearangos , I am not sure why the nose has that kind of artifacts, there can be many factors. For getting better idea about your results, I think you should test with more diverse images and clothes. Also, preparing the custom images similar to the viton dataset images may provide better results. Thanks.

soumyajit123dev commented 3 years ago

@josearangos I am also getting same error which you got. Please see error below

(label.cpu().numpy() == 11).astype(np.float)).cuda() RuntimeError: cuda runtime error (59) : device-side assert triggered at /pytorch/aten/src/THC/generic/THCTensorCopy.cpp:70

Please help me how to solve this? I hope you already solved it.

I generated the segmentation map from "Self Correction Human Parsing for ACGPN" repository (https://github.com/armheb/Self-Correction-Human-Parsing-for-ACGPN) and then converted the coloured output to gray scale. However I am getting the above error when I run test.py

Thanks for helping