Using CMU's OpenPose with facial and hands keypoints

akanazawa / hmr

Project page for End-to-end Recovery of Human Shape and Pose

Other

1.54k stars 395 forks source link

Using CMU's OpenPose with facial and hands keypoints #143

Closed jaygit123 closed 3 years ago

jaygit123 commented 4 years ago

Hi, Thanks for the code. I need facial expressions and hands / fingers as well in the resultant bvh file. So, I am looking to use CMU's OpenPose (instead of keras_Realtime_Multi-Person_Pose_Estimation).

Has anyone tried with CMU OpenPose and succeeded in including facial expressions and hands/fingers motions as well in the bvh? Do let me know.

jaygit123 commented 4 years ago

Hi @akanazawa, Could you help with the above request please?

Thanks.

akanazawa commented 3 years ago

Yes if you have more keypoints you can setup new terms in the reprojection loss and train with them. See SMPLify-X: https://github.com/vchoutas/smplify-x that goes in this direction.

Yes you need to write them out to frames.

Best,