I’m using the Nuitrack SDK on Unity for an application targeting the Arabic market.
Many of my users wear burqas, thobes, or abayas, which completely ruins the detection.
I’d like to know if there’s any way for me to fine-tune a MediaPipe model and replace the underlying detection engine. If not, is there at least a way to override the skeleton detection with a model running in parallel, so I can still use most of the great tooling for Unity but with my own pose detection engine?
Thanks!
Dear @joaopimenta,
Sent the answer to your question to your email.
Looking forward to hearing back from you there.
Thanks.
Other than communicating directly in email, information for anyone else who’re interested - there is a similar topic Is possible to integrate custom deep learning model into the Sdk?
We don’t have any official plans on this yet, but
(1) if you feel like you have a really good model and
(2) need to use Nuitrack at the same time (for some reason 
) and
(3) have a commercial project which depends on this, so you suppose to have significant licensing volume (100+ sensors)
feel free to contact us directly.