Matiur Rahman Minar

Please try to follow this guidelines: https://github.com/minar09/cp-vton-plus#testing-with-custom-images

Hi, @hfarhidzadeh , normally model should be training on multiple gpus if you set the id values correctly, by dividing the data according to the batch_size//no_of_gpus.

Hi @chethan-reddy2 , can you please provide more details about your input and input channels, e.g., segmentation type and labels, pose keypoints type and numbers etc.

Hi @arvind-iyer , this code is very heuristic, sure you can update the arguments based on your data. For example, in the case of the VITON dataset, they don't have...

@zandls-lovelulu , can you please clarify your step, and your input and output? More details will be appreciated. Thanks.

Please read this carefully: https://github.com/minar09/cp-vton-plus#data-preparation And this: https://github.com/minar09/cp-vton-plus#testing-with-custom-images Also, read our paper to understand the reason.

Defaults paths are already corrected. Just download the dataset and run the script.

Hi @vasujoshi111 , is the detectron2 joints ordering are same as openose? If not, you need to generate poses for your images with openpose coco-18 model.

Hi @vasujoshi111 , your results are obviously not as expected. Some people previously had sorting errors, but this one looks different. Please debug a little to see whether all the...

I think VITON dataset segmentation is generated with this: https://github.com/Engineering-Course/LIP_SSL Please check the original paper for more details.