[sadhorse]

YOLOA model has knowledge for dozens of objects. At every frame it is computing lots of parameters not related to your case. This is very wasteful and the main reason you can't achieve your performance goals.

[atoav]

I figured so much, any idea of models that are just person detecting?

[nasir]

So I guess what I saw is to use a big model to do annotation and develop a specialised smaller model.

[ekabod]

You can train a YOLO model from scratch.

[atoav]

Already checked that and have some data for it, just hoped getting people positions is such a typical usecase someone would have done a better model than I could here