search

hongsukchoi / 3DCrowdNet_RELEASE

Official Pytorch implementation of "Learning to Estimate Robust 3D Human Mesh from In-the-Wild Crowded Scenes", CVPR 2022
MIT License
155 stars 15 forks source link

About MuPoTs evaluation #5

Closed uyoung-jeong closed 2 years ago

uyoung-jeong commented 2 years ago

Your paper reports evaluation resluts of MuPoTS-3D, but there are several procedures I could not fully understand. 1) Your data loading script(data/MuPoTs/MuPoTs.py) reads preprocessed 2d pose estimator results based on annotation id. Does this mean that you used ground-truth bounding boxes for retrieving 2d predictions? 2) At https://github.com/hongsukchoi/3DCrowdNet_RELEASE/blob/c4d32c0dbb2675b022f9bc2848a39242d1142244/data/MuPoTs/MuPoTs.py#L80, your code specifies to use joint regressor weights of mpi-inf-3dhp dataset. Is this file made by yourself? As far as I know, previous works(including https://github.com/JiangWenPL/multiperson) used J_regressor_extra.npy for evaluation on mpi-inf-3dhp and MuPoTS.

hongsukchoi commented 2 years ago

1) No. 2) I got the file from my colleague. You can see the difference by visualizing keypoints in images. I am not sure right now but I think I used it because it has more accurate hip joints comparing to the GT.

uyoung-jeong commented 2 years ago

Thanks for the reply. :)

xljh0520 commented 1 year ago

Your paper reports evaluation resluts of MuPoTS-3D, but there are several procedures I could not fully understand. 1) Your data loading script(data/MuPoTs/MuPoTs.py) reads preprocessed 2d pose estimator results based on annotation id. Does this mean that you used ground-truth bounding boxes for retrieving 2d predictions? 2) At

https://github.com/hongsukchoi/3DCrowdNet_RELEASE/blob/c4d32c0dbb2675b022f9bc2848a39242d1142244/data/MuPoTs/MuPoTs.py#L80

, your code specifies to use joint regressor weights of mpi-inf-3dhp dataset. Is this file made by yourself? As far as I know, previous works(including https://github.com/JiangWenPL/multiperson) used J_regressor_extra.npy for evaluation on mpi-inf-3dhp and MuPoTS.

Hi, did you reproduce the result on MuPoTS? If so, could you share your preprocessed annotation file? Thanks in advance!

uyoung-jeong commented 1 year ago

Your paper reports evaluation resluts of MuPoTS-3D, but there are several procedures I could not fully understand. 1) Your data loading script(data/MuPoTs/MuPoTs.py) reads preprocessed 2d pose estimator results based on annotation id. Does this mean that you used ground-truth bounding boxes for retrieving 2d predictions? 2) At https://github.com/hongsukchoi/3DCrowdNet_RELEASE/blob/c4d32c0dbb2675b022f9bc2848a39242d1142244/data/MuPoTs/MuPoTs.py#L80

, your code specifies to use joint regressor weights of mpi-inf-3dhp dataset. Is this file made by yourself? As far as I know, previous works(including https://github.com/JiangWenPL/multiperson) used J_regressor_extra.npy for evaluation on mpi-inf-3dhp and MuPoTS.

Hi, did you reproduce the result on MuPoTS? If so, could you share your preprocessed annotation file? Thanks in advance!

No, I have no idea how to reproduce the result. It's better to ask the author to release related information.