2015/01/25 by Sourav Garg, Swagat Kumar, Garg, Sourav +5
Computer Science · Mathematics · #Advanced Vision and Imaging #Artificial intelligence #Closeness #Computer Vision and Pattern Recognition (cs.CV) #Computer science #Computer vision #FOS: Computer and information sciences #Frame (networking) #Human Pose and Action Recognition #Matching (statistics) #Mathematics #Monocular #Scheme (mathematics) #Tracking (education) #Video Surveillance and Tracking Methods #cs.CV
paper · pdf · doi:10.48550/arxiv.1501.06129
published in arXiv (Cornell University) (Cornell University) · 8 pages
arxiv created 2015/01/25 · openalex publication_date 2015/01/25 · arxiv updated 2015/01/27 · openalex created_date 2022/10/04 · openalex updated_date 2026/08/05
This paper looks into the problem of pedestrian tracking using a monocular,\npotentially moving, uncalibrated camera. The pedestrians are located in each\nframe using a standard human detector, which are then tracked in subsequent\nframes. This is a challenging problem as one has to deal with complex\nsituations like changing background, partial or full occlusion and camera\nmotion. In order to carry out successful tracking, it is necessary to resolve\nassociations between the detected windows in the current frame with those\nobtained from the previous frame. Compared to methods that use temporal windows\nincorporating past as well as future information, we attempt to make decision\non a frame-by-frame basis. An occlusion reasoning scheme is proposed to resolve\nthe association problem between a pair of consecutive frames by using an\naffinity matrix that defines the closeness between a pair of windows and then,\nuses a binary integer programming to obtain unique association between them. A\nsecond stage of verification based on SURF matching is used to deal with those\ncases where the above optimization scheme might yield wrong associations. The\nefficacy of the approach is demonstrated through experiments on several\nstandard pedestrian datasets.\n