I've been doing a search for tracking systems based on bounding box for several days (like Sort, DeepSort, Motpy, ...). But neither is nonrealtime.
My problem is based on the tracking of multiple objects in a video that is not in real time, my idea is to first process all the video making the detections and then apply the tracking (assuming that it is able to see the past, present and future) .
Also, I dont discard use features like DeepSort, to do the matchings.
But I can't find any algorithm that has nonrealtime tracking (that is, they only use past and present information).
My interest is to improve the tracking as much as possible.