Discriminative hierarchical part-based models for human parsing and action recognition

Yan Wang, Duan Tran, Zicheng Liao, David Forsyth

Research output: Contribution to journalArticle

Abstract

We consider the problem of parsing human poses and recognizing their actions in static images with part-based models. Most previous work in part-based models only considers rigid parts (e.g., torso, head, half limbs) guided by human anatomy. We argue that this representation of parts is not necessarily appropriate. In this paper, we introduce hierarchical poselets-a new representation for modeling the pose configuration of human bodies. Hierarchical poselets can be rigid parts, but they can also be parts that cover large portions of human bodies (e.g., torso + left arm). In the extreme case, they can be the whole bodies. The hierarchical poselets are organized in a hierarchical way via a structured model. Human parsing can be achieved by inferring the optimal labeling of this hierarchical model. The pose information captured by this hierarchical model can also be used as a intermediate representation for other high-level tasks. We demonstrate it in action recognition from static images.

Original languageEnglish (US)
Pages (from-to)3075-3102
Number of pages28
JournalJournal of Machine Learning Research
Volume13
StatePublished - Oct 1 2012

Fingerprint

Action Recognition
Parsing
Hierarchical Model
Model
Anatomy
Labeling
Extremes
Human
Cover
Configuration
Modeling
Demonstrate

Keywords

  • Action recognition
  • Hierarchical poselets
  • Human parsing
  • Maxmargin structured learning
  • Part-based models

ASJC Scopus subject areas

  • Software
  • Control and Systems Engineering
  • Statistics and Probability
  • Artificial Intelligence

Cite this

Discriminative hierarchical part-based models for human parsing and action recognition. / Wang, Yan; Tran, Duan; Liao, Zicheng; Forsyth, David.

In: Journal of Machine Learning Research, Vol. 13, 01.10.2012, p. 3075-3102.

Research output: Contribution to journalArticle

@article{cb887e1b4da34681a1281089541362c4,
title = "Discriminative hierarchical part-based models for human parsing and action recognition",
abstract = "We consider the problem of parsing human poses and recognizing their actions in static images with part-based models. Most previous work in part-based models only considers rigid parts (e.g., torso, head, half limbs) guided by human anatomy. We argue that this representation of parts is not necessarily appropriate. In this paper, we introduce hierarchical poselets-a new representation for modeling the pose configuration of human bodies. Hierarchical poselets can be rigid parts, but they can also be parts that cover large portions of human bodies (e.g., torso + left arm). In the extreme case, they can be the whole bodies. The hierarchical poselets are organized in a hierarchical way via a structured model. Human parsing can be achieved by inferring the optimal labeling of this hierarchical model. The pose information captured by this hierarchical model can also be used as a intermediate representation for other high-level tasks. We demonstrate it in action recognition from static images.",
keywords = "Action recognition, Hierarchical poselets, Human parsing, Maxmargin structured learning, Part-based models",
author = "Yan Wang and Duan Tran and Zicheng Liao and David Forsyth",
year = "2012",
month = "10",
day = "1",
language = "English (US)",
volume = "13",
pages = "3075--3102",
journal = "Journal of Machine Learning Research",
issn = "1532-4435",
publisher = "Microtome Publishing",

}

TY - JOUR

T1 - Discriminative hierarchical part-based models for human parsing and action recognition

AU - Wang, Yan

AU - Tran, Duan

AU - Liao, Zicheng

AU - Forsyth, David

PY - 2012/10/1

Y1 - 2012/10/1

N2 - We consider the problem of parsing human poses and recognizing their actions in static images with part-based models. Most previous work in part-based models only considers rigid parts (e.g., torso, head, half limbs) guided by human anatomy. We argue that this representation of parts is not necessarily appropriate. In this paper, we introduce hierarchical poselets-a new representation for modeling the pose configuration of human bodies. Hierarchical poselets can be rigid parts, but they can also be parts that cover large portions of human bodies (e.g., torso + left arm). In the extreme case, they can be the whole bodies. The hierarchical poselets are organized in a hierarchical way via a structured model. Human parsing can be achieved by inferring the optimal labeling of this hierarchical model. The pose information captured by this hierarchical model can also be used as a intermediate representation for other high-level tasks. We demonstrate it in action recognition from static images.

AB - We consider the problem of parsing human poses and recognizing their actions in static images with part-based models. Most previous work in part-based models only considers rigid parts (e.g., torso, head, half limbs) guided by human anatomy. We argue that this representation of parts is not necessarily appropriate. In this paper, we introduce hierarchical poselets-a new representation for modeling the pose configuration of human bodies. Hierarchical poselets can be rigid parts, but they can also be parts that cover large portions of human bodies (e.g., torso + left arm). In the extreme case, they can be the whole bodies. The hierarchical poselets are organized in a hierarchical way via a structured model. Human parsing can be achieved by inferring the optimal labeling of this hierarchical model. The pose information captured by this hierarchical model can also be used as a intermediate representation for other high-level tasks. We demonstrate it in action recognition from static images.

KW - Action recognition

KW - Hierarchical poselets

KW - Human parsing

KW - Maxmargin structured learning

KW - Part-based models

UR - http://www.scopus.com/inward/record.url?scp=84867855463&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84867855463&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:84867855463

VL - 13

SP - 3075

EP - 3102

JO - Journal of Machine Learning Research

JF - Journal of Machine Learning Research

SN - 1532-4435

ER -