How do you create the hand MANO ground truth and is there are code?

[spk921]

Hello,

Thank you for the work.
Is there are code that creating the ground truth of MANO hand annotation [pose , shape parameter] ?
I saw some article which use SGD to fit iteratively .. wondering is there are implementation of that.
I know you are busy but it would be really helpful if you can help that out.
Thank you.

[hassony2]

Hi !

Sorry for the very late answer :)
It all depends on what you are trying to fit to (2D joint locations, point clouds, depth map, ...)
In any case you are right that some form of MANO optimization can be implemented for that purpose.
You can for instance check https://github.com/shreyashampali/HOnnotate? where such fitting is performed for joint hand and object pose estimation from RGB-D data.

Best,

Yana