Matiur Rahman Minar

Hi @ngadminq , what is your application, could you please provide specific details? Thanks.

You can make parsing using the CIHP-PGN model, that is sufficient.

Hi @Quang-elec44 , sorry for my late response. I am sorry to say that I haven't been working on this repo for a long time. Also, this repository could be...

@linux-devil you can try from https://github.com/Engineering-Course/CIHP_PGN, https://github.com/Gaoyiminggithub/Graphonomy, https://github.com/prismformore/Multi-Task-Transformer, or https://github.com/facebookresearch/segment-anything . There might be many more.

Hi @josearangos , I am not sure whether it will work but you can try uncommenting this line https://github.com/minar09/LIP-JPPNet-TensorFlow/blob/ce669ad73daab753b1b664b07b87bd89cb175bbe/evaluate_parsing_JPPNet-s2.py#L9 in to something like this: `os.environ["CUDA_VISIBLE_DEVICES"] = "0,1"`

If you can generate the result, I am not sure if the environment has something to do with changing anything. Maybe you can debug your input/output more closely with the...

Hi @kira5511 , it seems your pretrained model loading is failed, that's why the predictions are incorrect. Can you please check whether you downloaded the pretrained model file into the...

Hi, @anantaarora , are you running on CPU? Using GPU would make faster maybe. Or you can refer to the original repository https://github.com/Engineering-Course/LIP_JPPNet. Thanks.

Hi @gamenerd457 , I haven't tried this on TensorFlow 2, however, the conversion from TensorFlow 1 to 2 should not be complex.

Hi @UmmaSaimaRahman , you can generate a lip_val_set.csv for your custom dataset following the script `create_heatmaps.py` in the `datasets/lip` folder. Hope that helps. Thanks.