algorithm - high performance object trackers in computer vision -
i'm interested in tracking multiple objects in video using (potentially moving) single camera in environment partial occlusions , variable distance objects camera. wondering of tracking algorithms can solve problem? experimented mean , cam-shift background subtraction wasn't able results.
i recommend reading through results of vot challenge 15. , continue researching selected trackers. http://www.votchallenge.net/vot2015/
correlation filter based trackers (like srdcf - https://www.cvl.isy.liu.se/research/objrec/visualtracking/regvistrack/ ) achieve high speed (due using fast fourier transform), state-of the-art results , not hard implement, first. (srdcf original implementation single-target, easy extend running multiple instances)
Comments
Post a Comment