Interacting Multiview Tracker

Ju Hong Yoon, Ming Hsuan Yang, Kuk Jin Yoon

Research output: Contribution to journalArticle

26 Citations (Scopus)

Abstract

A robust algorithm is proposed for tracking a target object in dynamic conditions including motion blurs, illumination changes, pose variations, and occlusions. To cope with these challenging factors, multiple trackers based on different feature representations are integrated within a probabilistic framework. Each view of the proposed multiview (multi-channel) feature learning algorithm is concerned with one particular feature representation of a target object from which a tracker is developed with different levels of reliability. With the multiple trackers, the proposed algorithm exploits tracker interaction and selection for robust tracking performance. In the tracker interaction, a transition probability matrix is used to estimate dependencies between trackers. Multiple trackers communicate with each other by sharing information of sample distributions. The tracker selection process determines the most reliable tracker with the highest probability. To account for object appearance changes, the transition probability matrix and tracker probability are updated in a recursive Bayesian framework by reflecting the tracker reliability measured by a robust tracker likelihood function that learns to account for both transient and stable appearance changes. Experimental results on benchmark datasets demonstrate that the proposed interacting multiview algorithm performs robustly and favorably against state-of-the-art methods in terms of several quantitative metrics.

Original languageEnglish
Article number7226831
Pages (from-to)903-917
Number of pages15
JournalIEEE transactions on pattern analysis and machine intelligence
Volume38
Issue number5
DOIs
Publication statusPublished - 2016 May 1

Fingerprint

Transition Probability Matrix
Motion Blur
Target
Robust Algorithm
Information Sharing
Likelihood Function
Interaction
Occlusion
Illumination
Learning Algorithm
Learning algorithms
Benchmark
Metric
Lighting
Experimental Results
Estimate
Demonstrate
Object
Framework

All Science Journal Classification (ASJC) codes

  • Software
  • Computer Vision and Pattern Recognition
  • Computational Theory and Mathematics
  • Artificial Intelligence
  • Applied Mathematics

Cite this

Yoon, Ju Hong ; Yang, Ming Hsuan ; Yoon, Kuk Jin. / Interacting Multiview Tracker. In: IEEE transactions on pattern analysis and machine intelligence. 2016 ; Vol. 38, No. 5. pp. 903-917.
@article{f6f2258212b04b60b09be4fab39c28e0,
title = "Interacting Multiview Tracker",
abstract = "A robust algorithm is proposed for tracking a target object in dynamic conditions including motion blurs, illumination changes, pose variations, and occlusions. To cope with these challenging factors, multiple trackers based on different feature representations are integrated within a probabilistic framework. Each view of the proposed multiview (multi-channel) feature learning algorithm is concerned with one particular feature representation of a target object from which a tracker is developed with different levels of reliability. With the multiple trackers, the proposed algorithm exploits tracker interaction and selection for robust tracking performance. In the tracker interaction, a transition probability matrix is used to estimate dependencies between trackers. Multiple trackers communicate with each other by sharing information of sample distributions. The tracker selection process determines the most reliable tracker with the highest probability. To account for object appearance changes, the transition probability matrix and tracker probability are updated in a recursive Bayesian framework by reflecting the tracker reliability measured by a robust tracker likelihood function that learns to account for both transient and stable appearance changes. Experimental results on benchmark datasets demonstrate that the proposed interacting multiview algorithm performs robustly and favorably against state-of-the-art methods in terms of several quantitative metrics.",
author = "Yoon, {Ju Hong} and Yang, {Ming Hsuan} and Yoon, {Kuk Jin}",
year = "2016",
month = "5",
day = "1",
doi = "10.1109/TPAMI.2015.2473862",
language = "English",
volume = "38",
pages = "903--917",
journal = "IEEE Transactions on Pattern Analysis and Machine Intelligence",
issn = "0162-8828",
publisher = "IEEE Computer Society",
number = "5",

}

Interacting Multiview Tracker. / Yoon, Ju Hong; Yang, Ming Hsuan; Yoon, Kuk Jin.

In: IEEE transactions on pattern analysis and machine intelligence, Vol. 38, No. 5, 7226831, 01.05.2016, p. 903-917.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Interacting Multiview Tracker

AU - Yoon, Ju Hong

AU - Yang, Ming Hsuan

AU - Yoon, Kuk Jin

PY - 2016/5/1

Y1 - 2016/5/1

N2 - A robust algorithm is proposed for tracking a target object in dynamic conditions including motion blurs, illumination changes, pose variations, and occlusions. To cope with these challenging factors, multiple trackers based on different feature representations are integrated within a probabilistic framework. Each view of the proposed multiview (multi-channel) feature learning algorithm is concerned with one particular feature representation of a target object from which a tracker is developed with different levels of reliability. With the multiple trackers, the proposed algorithm exploits tracker interaction and selection for robust tracking performance. In the tracker interaction, a transition probability matrix is used to estimate dependencies between trackers. Multiple trackers communicate with each other by sharing information of sample distributions. The tracker selection process determines the most reliable tracker with the highest probability. To account for object appearance changes, the transition probability matrix and tracker probability are updated in a recursive Bayesian framework by reflecting the tracker reliability measured by a robust tracker likelihood function that learns to account for both transient and stable appearance changes. Experimental results on benchmark datasets demonstrate that the proposed interacting multiview algorithm performs robustly and favorably against state-of-the-art methods in terms of several quantitative metrics.

AB - A robust algorithm is proposed for tracking a target object in dynamic conditions including motion blurs, illumination changes, pose variations, and occlusions. To cope with these challenging factors, multiple trackers based on different feature representations are integrated within a probabilistic framework. Each view of the proposed multiview (multi-channel) feature learning algorithm is concerned with one particular feature representation of a target object from which a tracker is developed with different levels of reliability. With the multiple trackers, the proposed algorithm exploits tracker interaction and selection for robust tracking performance. In the tracker interaction, a transition probability matrix is used to estimate dependencies between trackers. Multiple trackers communicate with each other by sharing information of sample distributions. The tracker selection process determines the most reliable tracker with the highest probability. To account for object appearance changes, the transition probability matrix and tracker probability are updated in a recursive Bayesian framework by reflecting the tracker reliability measured by a robust tracker likelihood function that learns to account for both transient and stable appearance changes. Experimental results on benchmark datasets demonstrate that the proposed interacting multiview algorithm performs robustly and favorably against state-of-the-art methods in terms of several quantitative metrics.

UR - http://www.scopus.com/inward/record.url?scp=84963828940&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84963828940&partnerID=8YFLogxK

U2 - 10.1109/TPAMI.2015.2473862

DO - 10.1109/TPAMI.2015.2473862

M3 - Article

AN - SCOPUS:84963828940

VL - 38

SP - 903

EP - 917

JO - IEEE Transactions on Pattern Analysis and Machine Intelligence

JF - IEEE Transactions on Pattern Analysis and Machine Intelligence

SN - 0162-8828

IS - 5

M1 - 7226831

ER -