View-invariant human activity recognition based on shape and motion features

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

47 Citations (Scopus)

Abstract

Recognizing human activities from image sequences is an active area of research in computer vision. Most of the previous work on activity recognition focuses on recognition from a single view and ignores the issue of view invariance. In this paper, we present a view invariant human activity recognition approach that uses both motion and shape information for activity representation. For each frame in the video, a 128 dimensional optical flow vector of the region of interest is used to represent the motion of the human body, and a 90 dimensional eigen-shape vector is used to represent the shape. Each activity is represented by a set of Hidden Markov Models (HMMs), where each model represents the activity from a different viewing direction, to realize view-invariance recognition. Experiments on a database of video clips of different activities show that our method is robust.

Original languageEnglish
Title of host publicationProceedings - IEEE Sixth International Symposium on Multimedia Software Engineering, MSE 2004
Pages546-556
Number of pages11
StatePublished - Dec 1 2004
EventProceedings - IEEE Sixth International Symposium on Multimedia Software Engineering, MSE 2004 - Miami, FL, United States
Duration: Dec 13 2004Dec 15 2004

Other

OtherProceedings - IEEE Sixth International Symposium on Multimedia Software Engineering, MSE 2004
CountryUnited States
CityMiami, FL
Period12/13/0412/15/04

Fingerprint

Invariance
Optical flows
Hidden Markov models
Computer vision
Experiments

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Niu, F., & Abdel-Mottaleb, M. (2004). View-invariant human activity recognition based on shape and motion features. In Proceedings - IEEE Sixth International Symposium on Multimedia Software Engineering, MSE 2004 (pp. 546-556)

View-invariant human activity recognition based on shape and motion features. / Niu, Feng; Abdel-Mottaleb, Mohamed.

Proceedings - IEEE Sixth International Symposium on Multimedia Software Engineering, MSE 2004. 2004. p. 546-556.

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

Niu, F & Abdel-Mottaleb, M 2004, View-invariant human activity recognition based on shape and motion features. in Proceedings - IEEE Sixth International Symposium on Multimedia Software Engineering, MSE 2004. pp. 546-556, Proceedings - IEEE Sixth International Symposium on Multimedia Software Engineering, MSE 2004, Miami, FL, United States, 12/13/04.
Niu F, Abdel-Mottaleb M. View-invariant human activity recognition based on shape and motion features. In Proceedings - IEEE Sixth International Symposium on Multimedia Software Engineering, MSE 2004. 2004. p. 546-556
Niu, Feng ; Abdel-Mottaleb, Mohamed. / View-invariant human activity recognition based on shape and motion features. Proceedings - IEEE Sixth International Symposium on Multimedia Software Engineering, MSE 2004. 2004. pp. 546-556
@inproceedings{94f715b7848f49a29b2716f2c73b1945,
title = "View-invariant human activity recognition based on shape and motion features",
abstract = "Recognizing human activities from image sequences is an active area of research in computer vision. Most of the previous work on activity recognition focuses on recognition from a single view and ignores the issue of view invariance. In this paper, we present a view invariant human activity recognition approach that uses both motion and shape information for activity representation. For each frame in the video, a 128 dimensional optical flow vector of the region of interest is used to represent the motion of the human body, and a 90 dimensional eigen-shape vector is used to represent the shape. Each activity is represented by a set of Hidden Markov Models (HMMs), where each model represents the activity from a different viewing direction, to realize view-invariance recognition. Experiments on a database of video clips of different activities show that our method is robust.",
author = "Feng Niu and Mohamed Abdel-Mottaleb",
year = "2004",
month = "12",
day = "1",
language = "English",
isbn = "0769522173",
pages = "546--556",
booktitle = "Proceedings - IEEE Sixth International Symposium on Multimedia Software Engineering, MSE 2004",

}

TY - GEN

T1 - View-invariant human activity recognition based on shape and motion features

AU - Niu, Feng

AU - Abdel-Mottaleb, Mohamed

PY - 2004/12/1

Y1 - 2004/12/1

N2 - Recognizing human activities from image sequences is an active area of research in computer vision. Most of the previous work on activity recognition focuses on recognition from a single view and ignores the issue of view invariance. In this paper, we present a view invariant human activity recognition approach that uses both motion and shape information for activity representation. For each frame in the video, a 128 dimensional optical flow vector of the region of interest is used to represent the motion of the human body, and a 90 dimensional eigen-shape vector is used to represent the shape. Each activity is represented by a set of Hidden Markov Models (HMMs), where each model represents the activity from a different viewing direction, to realize view-invariance recognition. Experiments on a database of video clips of different activities show that our method is robust.

AB - Recognizing human activities from image sequences is an active area of research in computer vision. Most of the previous work on activity recognition focuses on recognition from a single view and ignores the issue of view invariance. In this paper, we present a view invariant human activity recognition approach that uses both motion and shape information for activity representation. For each frame in the video, a 128 dimensional optical flow vector of the region of interest is used to represent the motion of the human body, and a 90 dimensional eigen-shape vector is used to represent the shape. Each activity is represented by a set of Hidden Markov Models (HMMs), where each model represents the activity from a different viewing direction, to realize view-invariance recognition. Experiments on a database of video clips of different activities show that our method is robust.

UR - http://www.scopus.com/inward/record.url?scp=20844450409&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=20844450409&partnerID=8YFLogxK

M3 - Conference contribution

SN - 0769522173

SN - 9780769522173

SP - 546

EP - 556

BT - Proceedings - IEEE Sixth International Symposium on Multimedia Software Engineering, MSE 2004

ER -