Some suggestion

[PrideOfHiigara]

Thank you for providing such an innovative model and approach. Based on actual usage, I sincerely offer some suggestions and look forward to your improvements in the model and updates:

1:I found that many seemingly normal hands are still not being detected. I feel that this area could benefit from adding some bad hand training data to improve recognition rates.
2:When two hands overlap, one hand is often not detected, which causes the non-overlapping hand to become deformed after redrawing.
3:The redrawing area may appear incongruous on the original image, especially with complex backgrounds. It feels like feathering could be added to address this issue.

[wenquanlu]

Thanks so much for your suggestions!
Regarding the point 1: the hand reconstruction and detection models are not trained in this project, they are mainly off-the-shelf. That is why I encourage others to contribute to this project by adding more detection and reconstruction models. There are plenty of them and each have different strength and capability.
Regarding the point 2: yes, overlapping hand is a limitation of current design, and it is also a limitation of hand reconstruction models from research perspective.
Regarding the point 3: OK, cool, thanks, yea I have seen some good results on Internet produced by using feathered masks.