Charting-based subspace learning for video-based human action classification

Vijay John (Lead / Corresponding author), Emanuele Trucco

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

We use charting, a non-linear dimensionality reduction algorithm, for articulated human motion classification in multi-view sequences or 3D data. Charting estimates automatically the intrinsic dimensionality of the latent subspace and preserves local neighbourhood and global structure of high-dimensional data. We classify human actions sub-sequences of varying lengths of skeletal poses, adopting a multi-layered subspace classification scheme with layered pruning and search. The sub-sequences of varying lengths of skeletal poses can be extracted using either markerless articulated tracking algorithms or markerless motion capture systems. We present a qualitative and quantitative comparison of single-subspace and multiple-subspace classification algorithms. We also identify the minimum length of action skeletal poses, required for accurate classification, using competing classification systems as the baseline. We test our motion classification framework on HumanEva, CMU, HDM05 and ACCAD mocap datasets and achieve similar or better classification accuracy than various comparable systems.
Original languageEnglish
Pages (from-to)119-132
Number of pages14
JournalMachine Vision and Applications
Volume25
Issue number1
Early online date14 May 2013
DOIs
Publication statusPublished - Jan 2014

Cite this

@article{98647c59a7f14eb9863d1549f5f584d5,
title = "Charting-based subspace learning for video-based human action classification",
abstract = "We use charting, a non-linear dimensionality reduction algorithm, for articulated human motion classification in multi-view sequences or 3D data. Charting estimates automatically the intrinsic dimensionality of the latent subspace and preserves local neighbourhood and global structure of high-dimensional data. We classify human actions sub-sequences of varying lengths of skeletal poses, adopting a multi-layered subspace classification scheme with layered pruning and search. The sub-sequences of varying lengths of skeletal poses can be extracted using either markerless articulated tracking algorithms or markerless motion capture systems. We present a qualitative and quantitative comparison of single-subspace and multiple-subspace classification algorithms. We also identify the minimum length of action skeletal poses, required for accurate classification, using competing classification systems as the baseline. We test our motion classification framework on HumanEva, CMU, HDM05 and ACCAD mocap datasets and achieve similar or better classification accuracy than various comparable systems.",
author = "Vijay John and Emanuele Trucco",
year = "2014",
month = "1",
doi = "10.1007/s00138-013-0508-y",
language = "English",
volume = "25",
pages = "119--132",
journal = "Machine Vision and Applications",
issn = "0932-8092",
publisher = "Springer Verlag",
number = "1",

}

Charting-based subspace learning for video-based human action classification. / John, Vijay (Lead / Corresponding author); Trucco, Emanuele.

In: Machine Vision and Applications, Vol. 25, No. 1, 01.2014, p. 119-132.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Charting-based subspace learning for video-based human action classification

AU - John, Vijay

AU - Trucco, Emanuele

PY - 2014/1

Y1 - 2014/1

N2 - We use charting, a non-linear dimensionality reduction algorithm, for articulated human motion classification in multi-view sequences or 3D data. Charting estimates automatically the intrinsic dimensionality of the latent subspace and preserves local neighbourhood and global structure of high-dimensional data. We classify human actions sub-sequences of varying lengths of skeletal poses, adopting a multi-layered subspace classification scheme with layered pruning and search. The sub-sequences of varying lengths of skeletal poses can be extracted using either markerless articulated tracking algorithms or markerless motion capture systems. We present a qualitative and quantitative comparison of single-subspace and multiple-subspace classification algorithms. We also identify the minimum length of action skeletal poses, required for accurate classification, using competing classification systems as the baseline. We test our motion classification framework on HumanEva, CMU, HDM05 and ACCAD mocap datasets and achieve similar or better classification accuracy than various comparable systems.

AB - We use charting, a non-linear dimensionality reduction algorithm, for articulated human motion classification in multi-view sequences or 3D data. Charting estimates automatically the intrinsic dimensionality of the latent subspace and preserves local neighbourhood and global structure of high-dimensional data. We classify human actions sub-sequences of varying lengths of skeletal poses, adopting a multi-layered subspace classification scheme with layered pruning and search. The sub-sequences of varying lengths of skeletal poses can be extracted using either markerless articulated tracking algorithms or markerless motion capture systems. We present a qualitative and quantitative comparison of single-subspace and multiple-subspace classification algorithms. We also identify the minimum length of action skeletal poses, required for accurate classification, using competing classification systems as the baseline. We test our motion classification framework on HumanEva, CMU, HDM05 and ACCAD mocap datasets and achieve similar or better classification accuracy than various comparable systems.

UR - http://www.scopus.com/inward/record.url?scp=84877261033&partnerID=8YFLogxK

U2 - 10.1007/s00138-013-0508-y

DO - 10.1007/s00138-013-0508-y

M3 - Article

VL - 25

SP - 119

EP - 132

JO - Machine Vision and Applications

JF - Machine Vision and Applications

SN - 0932-8092

IS - 1

ER -