A method for human action recognition

Research output: Contribution to journalArticle

98 Citations (Scopus)

Abstract

This article deals with the problem of classification of human activities from video. Our approach uses motion features that are computed very efficiently, and subsequently projected into a lower dimensional space where matching is performed. Each action is represented as a manifold in this lower dimensional space and matching is done by comparing these manifolds. To demonstrate the effectiveness of this approach, it was used on a large data set of similar actions, each performed by many different actors. Classification results were very accurate and show that this approach is robust to challenges such as variations in performers' physical attributes, color of clothing, and style of motion. An important result of this article is that the recovery of the three-dimensional properties of a moving person, or even the two-dimensional tracking of the person's limbs need not precede action recognition.

Original languageEnglish (US)
Pages (from-to)729-743
Number of pages15
JournalImage and Vision Computing
Volume21
Issue number8
DOIs
StatePublished - Aug 1 2003

Fingerprint

Color
Recovery

Keywords

  • Articulated motion
  • Human tracking
  • Motion recognition

Cite this

A method for human action recognition. / Masoud, Osama; Papanikolopoulos, Nikolaos P.

In: Image and Vision Computing, Vol. 21, No. 8, 01.08.2003, p. 729-743.

Research output: Contribution to journalArticle

@article{257cca7d2ba54f2094172b8e1d1df330,
title = "A method for human action recognition",
abstract = "This article deals with the problem of classification of human activities from video. Our approach uses motion features that are computed very efficiently, and subsequently projected into a lower dimensional space where matching is performed. Each action is represented as a manifold in this lower dimensional space and matching is done by comparing these manifolds. To demonstrate the effectiveness of this approach, it was used on a large data set of similar actions, each performed by many different actors. Classification results were very accurate and show that this approach is robust to challenges such as variations in performers' physical attributes, color of clothing, and style of motion. An important result of this article is that the recovery of the three-dimensional properties of a moving person, or even the two-dimensional tracking of the person's limbs need not precede action recognition.",
keywords = "Articulated motion, Human tracking, Motion recognition",
author = "Osama Masoud and Papanikolopoulos, {Nikolaos P}",
year = "2003",
month = "8",
day = "1",
doi = "10.1016/S0262-8856(03)00068-4",
language = "English (US)",
volume = "21",
pages = "729--743",
journal = "Image and Vision Computing",
issn = "0262-8856",
publisher = "Elsevier Limited",
number = "8",

}

TY - JOUR

T1 - A method for human action recognition

AU - Masoud, Osama

AU - Papanikolopoulos, Nikolaos P

PY - 2003/8/1

Y1 - 2003/8/1

N2 - This article deals with the problem of classification of human activities from video. Our approach uses motion features that are computed very efficiently, and subsequently projected into a lower dimensional space where matching is performed. Each action is represented as a manifold in this lower dimensional space and matching is done by comparing these manifolds. To demonstrate the effectiveness of this approach, it was used on a large data set of similar actions, each performed by many different actors. Classification results were very accurate and show that this approach is robust to challenges such as variations in performers' physical attributes, color of clothing, and style of motion. An important result of this article is that the recovery of the three-dimensional properties of a moving person, or even the two-dimensional tracking of the person's limbs need not precede action recognition.

AB - This article deals with the problem of classification of human activities from video. Our approach uses motion features that are computed very efficiently, and subsequently projected into a lower dimensional space where matching is performed. Each action is represented as a manifold in this lower dimensional space and matching is done by comparing these manifolds. To demonstrate the effectiveness of this approach, it was used on a large data set of similar actions, each performed by many different actors. Classification results were very accurate and show that this approach is robust to challenges such as variations in performers' physical attributes, color of clothing, and style of motion. An important result of this article is that the recovery of the three-dimensional properties of a moving person, or even the two-dimensional tracking of the person's limbs need not precede action recognition.

KW - Articulated motion

KW - Human tracking

KW - Motion recognition

UR - http://www.scopus.com/inward/record.url?scp=0038729570&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0038729570&partnerID=8YFLogxK

U2 - 10.1016/S0262-8856(03)00068-4

DO - 10.1016/S0262-8856(03)00068-4

M3 - Article

VL - 21

SP - 729

EP - 743

JO - Image and Vision Computing

JF - Image and Vision Computing

SN - 0262-8856

IS - 8

ER -