track hypotheses for objects that are not in the frame for the whole video

[DariusCorp]

Good day,

I adapted your algorithm to fit a specific need to track objects detected with yolo. The problem is I am not getting first frame tracks because I suppose that after the objects gets out of frame, the score reduces dramatically. I was wondering if you factored this into the algorithm because I do not see a workaround even if I adjust the scores in the PARAMS file