Ability to fine-tune pose landmark detection model with model maker

[RoadieHouse]

MediaPipe Solution (you are using)

Poser Landmark detection

Programming language

Python

Are you willing to contribute it

Yes

Describe the feature and the current behaviour/state

Currently pose landmark detection cannot be fine-tuned thorugh the model maker compared to other mediapipe solutions

Will this change the current API? How?

No response

Who will benefit with this feature?

Creators using pose estimation for specific use cases instead of generic ones

Please specify the use cases for this feature

Working on a workout app that focuses on calisthenics. There are many movements and unusual poses that mediapipe pose landmark detection does not handle well (e.g. being upside down). Therefore it would be great to have the ability to use the pose landmark model as a great baseline model and fine tune it for this (and other) specific use cases.

Any Other info

No response

[kuaashish]

Hi @RoadieHouse,

Thank you for suggesting the extension of the pose landmark functionality. We are forwarding your request internally, and our team will prioritize its implementation based on discussions and demand. We have noticed multiple requests for this feature, which underscores its importance.

Thank you!!

[harhit22]

i need it too

[arielfikru]

Still wait for this

[qemtek]

Another use-case I would like to finetune as the original model underperforms (but is very promising!)

[mcdonasd1212]

This would be a useful feature!

[xywang01]

This feature would be super helpful!

[dimitryvinichenko]

I need it too, but it is 8 month since issue was opened, is there any way to do it now?