Model field particles with positional appearance learning for sports player tracking

buir.advisorAksoy, Selim
dc.contributor.authorBaysal, Sermetcan
dc.departmentDepartment of Computer Engineeringen_US
dc.descriptionCataloged from PDF version of article.en_US
dc.descriptionThesis (Ph.D.): Bilkent University, Department of Computer Engineering, İhsan Doğramacı Bilkent University, 2016.en_US
dc.descriptionIncludes bibliographical references (leaves 66-74).en_US
dc.description.abstractTracking multiple players is crucial to analyzing sports videos in real time. Yet, illumination variations, background clutter, frequent occlusions among players who look similar in low-resolution, and non-linear motion patterns of the targets make sports player tracking di cult. Particle- ltering based approaches have been utilized for their ability in tracking under occlusion and rapid motions. Unlike the common practice of choosing particles on targets, we introduce the notion of shared particles densely sampled at xed positions on the model eld. Likelihoods of being on di erent particles are calculated for the targets using the proposed combined appearance and motion model. After globally distributing particles among the tracks, particles are weighted using an appearance model with a player detection score, and the track locations are updated by the weighted combination of the particles. This enables encapsulating the interactions among the targets in the state-space model and tracking players through challenging occlusions. We further introduce collective motion model and positional appearance learning to recover lost players and detect identity switches among the tracks. The proposed algorithm is embedded into a real player tracking system. Complete steps of the system are described and the proposed approach is evaluated on large-scale video. Experimental results show that the proposed tracker performs better than standard particle ltering and the state-of-the-art single-object trackers by losing less number of tracks and preserving more identities. Moreover, the proposed approach achieves a higher tracking accuracy with lower error rates on a publicly available soccer tracking dataset when compared to the previous methods.en_US
dc.description.provenanceSubmitted by Betül Özen ( on 2016-07-12T12:25:49Z No. of bitstreams: 1 10114437.pdf: 25592320 bytes, checksum: c2592373324b13579d87f6324bc3dcab (MD5)en
dc.description.provenanceMade available in DSpace on 2016-07-12T12:25:49Z (GMT). No. of bitstreams: 1 10114437.pdf: 25592320 bytes, checksum: c2592373324b13579d87f6324bc3dcab (MD5) Previous issue date: 2016-06en
dc.description.statementofresponsibilityby Sermetcan Baysal.en_US
dc.format.extentxii, 74 leaves : illustrations, charts.en_US
dc.publisherBilkent Universityen_US
dc.subjectSports video analysisen_US
dc.subjectSports player trackingen_US
dc.subjectMultiple object trackingen_US
dc.subjectModel eld particlesen_US
dc.subjectPositional appearance learningen_US
dc.subjectCollective motion modelen_US
dc.titleModel field particles with positional appearance learning for sports player trackingen_US
dc.title.alternativeSporcu takibi için saha modeli parçacıkları ve pozisyon tabanlı görünüm öğrenimien_US
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
24.41 MB
Adobe Portable Document Format
Full printable version
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
1.71 KB
Item-specific license agreed upon to submission