Human activity recognition using overlapping multi-feature descriptor

S. Y. Cho, Hyeran Byun

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

An efficient overlapping multi-feature descriptor and classification scheme for human activity recognition is introduced. The descriptor is constructed by overlapping global feature combinations of multi-frames using a Hankel matrix representation. The descriptor captures the local and temporal information while overcoming the limitations of global features using an overlapping combination scheme. In addition, a random forests classifier is used to cope with noise in the descriptor that can be obtained from no-activity frames in a video. Using this framework, it is shown that the approach outperforms the state-of-the-art methods using the KTH dataset and a much more complex human interaction dataset.

Original languageEnglish
Pages (from-to)1275-1277
Number of pages3
JournalElectronics Letters
Volume47
Issue number23
DOIs
Publication statusPublished - 2011 Nov 10

Fingerprint

Classifiers

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering

Cite this

@article{c93b1f631ce745a283e3540a4528d345,
title = "Human activity recognition using overlapping multi-feature descriptor",
abstract = "An efficient overlapping multi-feature descriptor and classification scheme for human activity recognition is introduced. The descriptor is constructed by overlapping global feature combinations of multi-frames using a Hankel matrix representation. The descriptor captures the local and temporal information while overcoming the limitations of global features using an overlapping combination scheme. In addition, a random forests classifier is used to cope with noise in the descriptor that can be obtained from no-activity frames in a video. Using this framework, it is shown that the approach outperforms the state-of-the-art methods using the KTH dataset and a much more complex human interaction dataset.",
author = "Cho, {S. Y.} and Hyeran Byun",
year = "2011",
month = "11",
day = "10",
doi = "10.1049/el.2011.2550",
language = "English",
volume = "47",
pages = "1275--1277",
journal = "Electronics Letters",
issn = "0013-5194",
publisher = "Institution of Engineering and Technology",
number = "23",

}

Human activity recognition using overlapping multi-feature descriptor. / Cho, S. Y.; Byun, Hyeran.

In: Electronics Letters, Vol. 47, No. 23, 10.11.2011, p. 1275-1277.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Human activity recognition using overlapping multi-feature descriptor

AU - Cho, S. Y.

AU - Byun, Hyeran

PY - 2011/11/10

Y1 - 2011/11/10

N2 - An efficient overlapping multi-feature descriptor and classification scheme for human activity recognition is introduced. The descriptor is constructed by overlapping global feature combinations of multi-frames using a Hankel matrix representation. The descriptor captures the local and temporal information while overcoming the limitations of global features using an overlapping combination scheme. In addition, a random forests classifier is used to cope with noise in the descriptor that can be obtained from no-activity frames in a video. Using this framework, it is shown that the approach outperforms the state-of-the-art methods using the KTH dataset and a much more complex human interaction dataset.

AB - An efficient overlapping multi-feature descriptor and classification scheme for human activity recognition is introduced. The descriptor is constructed by overlapping global feature combinations of multi-frames using a Hankel matrix representation. The descriptor captures the local and temporal information while overcoming the limitations of global features using an overlapping combination scheme. In addition, a random forests classifier is used to cope with noise in the descriptor that can be obtained from no-activity frames in a video. Using this framework, it is shown that the approach outperforms the state-of-the-art methods using the KTH dataset and a much more complex human interaction dataset.

UR - http://www.scopus.com/inward/record.url?scp=81255184672&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=81255184672&partnerID=8YFLogxK

U2 - 10.1049/el.2011.2550

DO - 10.1049/el.2011.2550

M3 - Article

VL - 47

SP - 1275

EP - 1277

JO - Electronics Letters

JF - Electronics Letters

SN - 0013-5194

IS - 23

ER -