search

google-ai-edge / mediapipe

Cross-platform, customizable ML solutions for live and streaming media.
https://ai.google.dev/edge/mediapipe
Apache License 2.0
27.49k stars 5.15k forks source link

Holistic Key-point Modifications #3404

Closed Modyng closed 1 year ago

Modyng commented 2 years ago

System information (Please provide as much relevant information as possible) The Solution is HOLISTIC. Programming Language is Python Latest MediaPipe Version Os is Ubuntu (kernel do not matter)

Describe the behavior: MediaPipe Holistic has lots of keypoints for POSE, Hands and FaceMesh but i dont want all of the keypoins because they have some challenges or conflicts with other keypoints.

For example Holistic has two different key points on hands, including Pose ( which has 4 keypoints on each hand ) and Hand ( which has 21 keyPoints for each Hand). But those are not needed for me and they are somehow additional information which could be neglected or they are somehow redundant!( no need to be there)

The same happen for Face, Pose keypoints has some points on face while the Holistic faceMesh include all those points in the face. which could increase model complexity and overlapping.

Is there a way to remove some redundant ( additional ) keypoints from Holistic model in order to train just the usefull keypoints ?

Thanks Best regards

sureshdagooglecom commented 2 years ago

Hi @Modyng , It is possible to omit each of the 6 models from Holistic. Within each model, there is no easy way to reduce the set of landmarks predicted by the model.

Modyng commented 2 years ago

@sureshdagooglecom and @jiuqiant

Hi and Thanks for the reply,

i'm looking for the hard way! because i dont want to omit none of the 6 models but just some landmarks of each.

Would you please point me to that direction

Thanks best regards

Ared521 commented 2 years ago

Hello, I also have this problem. Have you solved it now? Please.

kuaashish commented 1 year ago

Hello @Modyng, We are upgrading the MediaPipe Legacy Solutions to new MediaPipe solutions However, the libraries, documentation, and source code for all the MediapPipe Legacy Solutions will continue to be available in our GitHub repository and through library distribution services, such as Maven and NPM.

You can continue to use those legacy solutions in your applications if you choose. Though, we would request you to check new MediaPipe solutions which can help you more easily build and customize ML solutions for your applications. These new solutions will provide a superset of capabilities available in the legacy solutions. Thank you

github-actions[bot] commented 1 year ago

This issue has been marked stale because it has no recent activity since 7 days. It will be closed if no further activity occurs. Thank you.

github-actions[bot] commented 1 year ago

This issue was closed due to lack of activity after being marked stale for past 7 days.

google-ml-butler[bot] commented 1 year ago

Are you satisfied with the resolution of your issue? Yes No