Multi-animal and multi-camera setup inquiry

[bolarinlaw]

I am implementing deepethogram to analyze behavior in a double decker cage setup with two Rats. I am concerned about labeling noise when two animals perform contradictory behaviors simultaneously (e.g., Rat A drinking while Rat B jumps).

Does the current architecture handle frame inputs with multiple animals well? Additionally, I have a top, right, and left camera angle available. Does deepethogram's multi-video feature support synced labeling or is it best to train each camera in a separate model.