Monocular tracking of 3D human motion with a coordinated mixture of factor analyzers

Rui Li, Ming Hsuan Yang, Stan Sclaroff, Tai Peng Tian

Research output: Chapter in Book/Report/Conference proceedingConference contribution

25 Citations (Scopus)

Abstract

Filtering bused algorithms have become popular in trucking human body pose. Such algorithms can suffer the curse of dimensionality due to the high dimensionality of the pose state space; therefore, efforts have been dedicated to either smart sampling or reducing the dimensionality of the original pose state space. In this paper, a novel formulation that employs a dimensionality reduced state space for multi-hypothesis tracking is proposed. During off-line training, a mixture of factor analyzers is learned. Each factor analyzer can be thought of as a "local dimensionality reducer" that, locally approximates the pose manifold. Global coordination between local factor analyzers is achieved by learning a set of linear mixture functions that enforces agreement between local factor analyzers. The formulation allows easy bidirectional mapping between the original body pose space and the low-dimensional space. During online tracking, the clusters of factor anlyzers are utilized in a multiple hypothesis tracking algorithm. Experiments demonstrate that the proposed algorithm tracks 3D body pose efficiently and accurately, even when self-occlusion, motion blur and large limb movements occur. Quantitative comparisons show that the formulation produces more accurate 3D pose estimates over time than those that can be obtained via a number of previously-proposed particle filtering based tracking algorithms.

Original languageEnglish
Title of host publicationComputer Vision - ECCV 2006, 9th European Conference on Computer Vision, Proceedings
PublisherSpringer Verlag
Pages137-150
Number of pages14
ISBN (Print)3540338349, 9783540338345
Publication statusPublished - 2006 Jan 1
Event9th European Conference on Computer Vision, ECCV 2006 - Graz, Austria
Duration: 2006 May 72006 May 13

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume3952 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference9th European Conference on Computer Vision, ECCV 2006
CountryAustria
CityGraz
Period06/5/706/5/13

Fingerprint

Dimensionality
Motion
State Space
Formulation
Motion Blur
Particle Filtering
Curse of Dimensionality
Occlusion
Filtering
Human
Sampling
Line
Estimate
Demonstrate
Experiment
Experiments

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Li, R., Yang, M. H., Sclaroff, S., & Tian, T. P. (2006). Monocular tracking of 3D human motion with a coordinated mixture of factor analyzers. In Computer Vision - ECCV 2006, 9th European Conference on Computer Vision, Proceedings (pp. 137-150). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3952 LNCS). Springer Verlag.
Li, Rui ; Yang, Ming Hsuan ; Sclaroff, Stan ; Tian, Tai Peng. / Monocular tracking of 3D human motion with a coordinated mixture of factor analyzers. Computer Vision - ECCV 2006, 9th European Conference on Computer Vision, Proceedings. Springer Verlag, 2006. pp. 137-150 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{15d5ea9f8dee447483c97eb3acb8b29f,
title = "Monocular tracking of 3D human motion with a coordinated mixture of factor analyzers",
abstract = "Filtering bused algorithms have become popular in trucking human body pose. Such algorithms can suffer the curse of dimensionality due to the high dimensionality of the pose state space; therefore, efforts have been dedicated to either smart sampling or reducing the dimensionality of the original pose state space. In this paper, a novel formulation that employs a dimensionality reduced state space for multi-hypothesis tracking is proposed. During off-line training, a mixture of factor analyzers is learned. Each factor analyzer can be thought of as a {"}local dimensionality reducer{"} that, locally approximates the pose manifold. Global coordination between local factor analyzers is achieved by learning a set of linear mixture functions that enforces agreement between local factor analyzers. The formulation allows easy bidirectional mapping between the original body pose space and the low-dimensional space. During online tracking, the clusters of factor anlyzers are utilized in a multiple hypothesis tracking algorithm. Experiments demonstrate that the proposed algorithm tracks 3D body pose efficiently and accurately, even when self-occlusion, motion blur and large limb movements occur. Quantitative comparisons show that the formulation produces more accurate 3D pose estimates over time than those that can be obtained via a number of previously-proposed particle filtering based tracking algorithms.",
author = "Rui Li and Yang, {Ming Hsuan} and Stan Sclaroff and Tian, {Tai Peng}",
year = "2006",
month = "1",
day = "1",
language = "English",
isbn = "3540338349",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "137--150",
booktitle = "Computer Vision - ECCV 2006, 9th European Conference on Computer Vision, Proceedings",
address = "Germany",

}

Li, R, Yang, MH, Sclaroff, S & Tian, TP 2006, Monocular tracking of 3D human motion with a coordinated mixture of factor analyzers. in Computer Vision - ECCV 2006, 9th European Conference on Computer Vision, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 3952 LNCS, Springer Verlag, pp. 137-150, 9th European Conference on Computer Vision, ECCV 2006, Graz, Austria, 06/5/7.

Monocular tracking of 3D human motion with a coordinated mixture of factor analyzers. / Li, Rui; Yang, Ming Hsuan; Sclaroff, Stan; Tian, Tai Peng.

Computer Vision - ECCV 2006, 9th European Conference on Computer Vision, Proceedings. Springer Verlag, 2006. p. 137-150 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3952 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Monocular tracking of 3D human motion with a coordinated mixture of factor analyzers

AU - Li, Rui

AU - Yang, Ming Hsuan

AU - Sclaroff, Stan

AU - Tian, Tai Peng

PY - 2006/1/1

Y1 - 2006/1/1

N2 - Filtering bused algorithms have become popular in trucking human body pose. Such algorithms can suffer the curse of dimensionality due to the high dimensionality of the pose state space; therefore, efforts have been dedicated to either smart sampling or reducing the dimensionality of the original pose state space. In this paper, a novel formulation that employs a dimensionality reduced state space for multi-hypothesis tracking is proposed. During off-line training, a mixture of factor analyzers is learned. Each factor analyzer can be thought of as a "local dimensionality reducer" that, locally approximates the pose manifold. Global coordination between local factor analyzers is achieved by learning a set of linear mixture functions that enforces agreement between local factor analyzers. The formulation allows easy bidirectional mapping between the original body pose space and the low-dimensional space. During online tracking, the clusters of factor anlyzers are utilized in a multiple hypothesis tracking algorithm. Experiments demonstrate that the proposed algorithm tracks 3D body pose efficiently and accurately, even when self-occlusion, motion blur and large limb movements occur. Quantitative comparisons show that the formulation produces more accurate 3D pose estimates over time than those that can be obtained via a number of previously-proposed particle filtering based tracking algorithms.

AB - Filtering bused algorithms have become popular in trucking human body pose. Such algorithms can suffer the curse of dimensionality due to the high dimensionality of the pose state space; therefore, efforts have been dedicated to either smart sampling or reducing the dimensionality of the original pose state space. In this paper, a novel formulation that employs a dimensionality reduced state space for multi-hypothesis tracking is proposed. During off-line training, a mixture of factor analyzers is learned. Each factor analyzer can be thought of as a "local dimensionality reducer" that, locally approximates the pose manifold. Global coordination between local factor analyzers is achieved by learning a set of linear mixture functions that enforces agreement between local factor analyzers. The formulation allows easy bidirectional mapping between the original body pose space and the low-dimensional space. During online tracking, the clusters of factor anlyzers are utilized in a multiple hypothesis tracking algorithm. Experiments demonstrate that the proposed algorithm tracks 3D body pose efficiently and accurately, even when self-occlusion, motion blur and large limb movements occur. Quantitative comparisons show that the formulation produces more accurate 3D pose estimates over time than those that can be obtained via a number of previously-proposed particle filtering based tracking algorithms.

UR - http://www.scopus.com/inward/record.url?scp=33745868742&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33745868742&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:33745868742

SN - 3540338349

SN - 9783540338345

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 137

EP - 150

BT - Computer Vision - ECCV 2006, 9th European Conference on Computer Vision, Proceedings

PB - Springer Verlag

ER -

Li R, Yang MH, Sclaroff S, Tian TP. Monocular tracking of 3D human motion with a coordinated mixture of factor analyzers. In Computer Vision - ECCV 2006, 9th European Conference on Computer Vision, Proceedings. Springer Verlag. 2006. p. 137-150. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).