Human pose tracking using online latent structured support vector machine

Kai Lung Hua, Irawati Nurmala Sari, Mei Chen Yeh

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

1 Citation (Scopus)

Abstract

Tracking human poses in a video is a challenging problem and has numerous applications. The task is particularly difficult in realistic scenes because of several intrinsic and extrinsic factors, including complicated and fast movements, occlusions and lighting changes. We propose an online learning approach for tracking human poses using latent structured Support Vector Machine (SVM). The first frame in a video is used for training, in which body parts are initialized by users and tracking models are learned using latent structured SVM. The models are updated for each subsequent frame in the video sequence. To solve the occlusion problem, we formulate a Prize-Collecting Steiner tree (PCST) problem and use a branch-and-cut algorithm to refine the detection of body parts. Experiments using several challenging videos demonstrate that the proposed method outperforms two state-of-the-art methods.

Original languageEnglish
Title of host publicationMultiMedia Modeling - 23rd International Conference, MMM 2017, Proceedings
EditorsGylfi Thór Gudmundsson, Shin’ichi Satoh, Laurent Amsaleg, Björn Thór Jónsson, Cathal Gurrin
PublisherSpringer Verlag
Pages626-637
Number of pages12
ISBN (Print)9783319518107
DOIs
Publication statusPublished - 2017 Jan 1
Event23rd International Conference on MultiMedia Modeling, MMM 2017 - Reykjavik, Iceland
Duration: 2017 Jan 42017 Jan 6

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10132 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other23rd International Conference on MultiMedia Modeling, MMM 2017
CountryIceland
CityReykjavik
Period17/1/417/1/6

Fingerprint

Support vector machines
Support Vector Machine
Occlusion
Steiner Tree Problem
Branch-and-cut
Online Learning
Lighting
Experiments
Model
Demonstrate
Experiment
Human
Movement
Training

Keywords

  • Body parts
  • Human pose tracking
  • Latent structured SVM
  • Online learning

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Hua, K. L., Sari, I. N., & Yeh, M. C. (2017). Human pose tracking using online latent structured support vector machine. In G. T. Gudmundsson, S. Satoh, L. Amsaleg, B. T. Jónsson, & C. Gurrin (Eds.), MultiMedia Modeling - 23rd International Conference, MMM 2017, Proceedings (pp. 626-637). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10132 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-51811-4_51

Human pose tracking using online latent structured support vector machine. / Hua, Kai Lung; Sari, Irawati Nurmala; Yeh, Mei Chen.

MultiMedia Modeling - 23rd International Conference, MMM 2017, Proceedings. ed. / Gylfi Thór Gudmundsson; Shin’ichi Satoh; Laurent Amsaleg; Björn Thór Jónsson; Cathal Gurrin. Springer Verlag, 2017. p. 626-637 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10132 LNCS).

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

Hua, KL, Sari, IN & Yeh, MC 2017, Human pose tracking using online latent structured support vector machine. in GT Gudmundsson, S Satoh, L Amsaleg, BT Jónsson & C Gurrin (eds), MultiMedia Modeling - 23rd International Conference, MMM 2017, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10132 LNCS, Springer Verlag, pp. 626-637, 23rd International Conference on MultiMedia Modeling, MMM 2017, Reykjavik, Iceland, 17/1/4. https://doi.org/10.1007/978-3-319-51811-4_51
Hua KL, Sari IN, Yeh MC. Human pose tracking using online latent structured support vector machine. In Gudmundsson GT, Satoh S, Amsaleg L, Jónsson BT, Gurrin C, editors, MultiMedia Modeling - 23rd International Conference, MMM 2017, Proceedings. Springer Verlag. 2017. p. 626-637. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-51811-4_51
Hua, Kai Lung ; Sari, Irawati Nurmala ; Yeh, Mei Chen. / Human pose tracking using online latent structured support vector machine. MultiMedia Modeling - 23rd International Conference, MMM 2017, Proceedings. editor / Gylfi Thór Gudmundsson ; Shin’ichi Satoh ; Laurent Amsaleg ; Björn Thór Jónsson ; Cathal Gurrin. Springer Verlag, 2017. pp. 626-637 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{73307cc48c4f4a3091c763bdcc7b1dd9,
title = "Human pose tracking using online latent structured support vector machine",
abstract = "Tracking human poses in a video is a challenging problem and has numerous applications. The task is particularly difficult in realistic scenes because of several intrinsic and extrinsic factors, including complicated and fast movements, occlusions and lighting changes. We propose an online learning approach for tracking human poses using latent structured Support Vector Machine (SVM). The first frame in a video is used for training, in which body parts are initialized by users and tracking models are learned using latent structured SVM. The models are updated for each subsequent frame in the video sequence. To solve the occlusion problem, we formulate a Prize-Collecting Steiner tree (PCST) problem and use a branch-and-cut algorithm to refine the detection of body parts. Experiments using several challenging videos demonstrate that the proposed method outperforms two state-of-the-art methods.",
keywords = "Body parts, Human pose tracking, Latent structured SVM, Online learning",
author = "Hua, {Kai Lung} and Sari, {Irawati Nurmala} and Yeh, {Mei Chen}",
year = "2017",
month = "1",
day = "1",
doi = "10.1007/978-3-319-51811-4_51",
language = "English",
isbn = "9783319518107",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "626--637",
editor = "Gudmundsson, {Gylfi Th{\'o}r} and Shin’ichi Satoh and Laurent Amsaleg and J{\'o}nsson, {Bj{\"o}rn Th{\'o}r} and Cathal Gurrin",
booktitle = "MultiMedia Modeling - 23rd International Conference, MMM 2017, Proceedings",

}

TY - GEN

T1 - Human pose tracking using online latent structured support vector machine

AU - Hua, Kai Lung

AU - Sari, Irawati Nurmala

AU - Yeh, Mei Chen

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Tracking human poses in a video is a challenging problem and has numerous applications. The task is particularly difficult in realistic scenes because of several intrinsic and extrinsic factors, including complicated and fast movements, occlusions and lighting changes. We propose an online learning approach for tracking human poses using latent structured Support Vector Machine (SVM). The first frame in a video is used for training, in which body parts are initialized by users and tracking models are learned using latent structured SVM. The models are updated for each subsequent frame in the video sequence. To solve the occlusion problem, we formulate a Prize-Collecting Steiner tree (PCST) problem and use a branch-and-cut algorithm to refine the detection of body parts. Experiments using several challenging videos demonstrate that the proposed method outperforms two state-of-the-art methods.

AB - Tracking human poses in a video is a challenging problem and has numerous applications. The task is particularly difficult in realistic scenes because of several intrinsic and extrinsic factors, including complicated and fast movements, occlusions and lighting changes. We propose an online learning approach for tracking human poses using latent structured Support Vector Machine (SVM). The first frame in a video is used for training, in which body parts are initialized by users and tracking models are learned using latent structured SVM. The models are updated for each subsequent frame in the video sequence. To solve the occlusion problem, we formulate a Prize-Collecting Steiner tree (PCST) problem and use a branch-and-cut algorithm to refine the detection of body parts. Experiments using several challenging videos demonstrate that the proposed method outperforms two state-of-the-art methods.

KW - Body parts

KW - Human pose tracking

KW - Latent structured SVM

KW - Online learning

UR - http://www.scopus.com/inward/record.url?scp=85009739445&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85009739445&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-51811-4_51

DO - 10.1007/978-3-319-51811-4_51

M3 - Conference contribution

AN - SCOPUS:85009739445

SN - 9783319518107

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

SP - 626

EP - 637

BT - MultiMedia Modeling - 23rd International Conference, MMM 2017, Proceedings

A2 - Gudmundsson, Gylfi Thór

A2 - Satoh, Shin’ichi

A2 - Amsaleg, Laurent

A2 - Jónsson, Björn Thór

A2 - Gurrin, Cathal

PB - Springer Verlag

ER -