Bio-inspired dynamic 3D discriminative skeletal features for human action recognition

Rizwan Chaudhry, Ferda Ofli, Gregorij Kurillo, Ruzena Bajcsy, Rene Vidal

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

90 Citations (Scopus)

Abstract

Over the last few years, with the immense popularity of the Kinect, there has been renewed interest in developing methods for human gesture and action recognition from 3D data. A number of approaches have been proposed that extract representative features from 3D depth data, a reconstructed 3D surface mesh or more commonly from the recovered estimate of the human skeleton. Recent advances in neuroscience have discovered a neural encoding of static 3D shapes in primate infero-temporal cortex that can be represented as a hierarchy of medial axis and surface features. We hypothesize a similar neural encoding might also exist for 3D shapes in motion and propose a hierarchy of dynamic medial axis structures at several spatio-temporal scales that can be modeled using a set of Linear Dynamical Systems (LDSs). We then propose novel discriminative metrics for comparing these sets of LDSs for the task of human activity recognition. Combined with simple classification frameworks, our proposed features and corresponding hierarchical dynamical models provide the highest human activity recognition rates as compared to state-of-the-art methods on several skeletal datasets.

Original languageEnglish
Title of host publicationIEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops
Pages471-478
Number of pages8
DOIs
Publication statusPublished - 8 Oct 2013
Externally publishedYes
Event2013 IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2013 - Portland, OR, United States
Duration: 23 Jun 201328 Jun 2013

Other

Other2013 IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2013
CountryUnited States
CityPortland, OR
Period23/6/1328/6/13

Fingerprint

Dynamical systems
Primates

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Electrical and Electronic Engineering

Cite this

Chaudhry, R., Ofli, F., Kurillo, G., Bajcsy, R., & Vidal, R. (2013). Bio-inspired dynamic 3D discriminative skeletal features for human action recognition. In IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops (pp. 471-478). [6595916] https://doi.org/10.1109/CVPRW.2013.153

Bio-inspired dynamic 3D discriminative skeletal features for human action recognition. / Chaudhry, Rizwan; Ofli, Ferda; Kurillo, Gregorij; Bajcsy, Ruzena; Vidal, Rene.

IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops. 2013. p. 471-478 6595916.

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

Chaudhry, R, Ofli, F, Kurillo, G, Bajcsy, R & Vidal, R 2013, Bio-inspired dynamic 3D discriminative skeletal features for human action recognition. in IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops., 6595916, pp. 471-478, 2013 IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2013, Portland, OR, United States, 23/6/13. https://doi.org/10.1109/CVPRW.2013.153
Chaudhry R, Ofli F, Kurillo G, Bajcsy R, Vidal R. Bio-inspired dynamic 3D discriminative skeletal features for human action recognition. In IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops. 2013. p. 471-478. 6595916 https://doi.org/10.1109/CVPRW.2013.153
Chaudhry, Rizwan ; Ofli, Ferda ; Kurillo, Gregorij ; Bajcsy, Ruzena ; Vidal, Rene. / Bio-inspired dynamic 3D discriminative skeletal features for human action recognition. IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops. 2013. pp. 471-478
@inproceedings{16ccd9c1830b4d8c81bc410adc96cd2c,
title = "Bio-inspired dynamic 3D discriminative skeletal features for human action recognition",
abstract = "Over the last few years, with the immense popularity of the Kinect, there has been renewed interest in developing methods for human gesture and action recognition from 3D data. A number of approaches have been proposed that extract representative features from 3D depth data, a reconstructed 3D surface mesh or more commonly from the recovered estimate of the human skeleton. Recent advances in neuroscience have discovered a neural encoding of static 3D shapes in primate infero-temporal cortex that can be represented as a hierarchy of medial axis and surface features. We hypothesize a similar neural encoding might also exist for 3D shapes in motion and propose a hierarchy of dynamic medial axis structures at several spatio-temporal scales that can be modeled using a set of Linear Dynamical Systems (LDSs). We then propose novel discriminative metrics for comparing these sets of LDSs for the task of human activity recognition. Combined with simple classification frameworks, our proposed features and corresponding hierarchical dynamical models provide the highest human activity recognition rates as compared to state-of-the-art methods on several skeletal datasets.",
author = "Rizwan Chaudhry and Ferda Ofli and Gregorij Kurillo and Ruzena Bajcsy and Rene Vidal",
year = "2013",
month = "10",
day = "8",
doi = "10.1109/CVPRW.2013.153",
language = "English",
isbn = "9780769549903",
pages = "471--478",
booktitle = "IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops",

}

TY - GEN

T1 - Bio-inspired dynamic 3D discriminative skeletal features for human action recognition

AU - Chaudhry, Rizwan

AU - Ofli, Ferda

AU - Kurillo, Gregorij

AU - Bajcsy, Ruzena

AU - Vidal, Rene

PY - 2013/10/8

Y1 - 2013/10/8

N2 - Over the last few years, with the immense popularity of the Kinect, there has been renewed interest in developing methods for human gesture and action recognition from 3D data. A number of approaches have been proposed that extract representative features from 3D depth data, a reconstructed 3D surface mesh or more commonly from the recovered estimate of the human skeleton. Recent advances in neuroscience have discovered a neural encoding of static 3D shapes in primate infero-temporal cortex that can be represented as a hierarchy of medial axis and surface features. We hypothesize a similar neural encoding might also exist for 3D shapes in motion and propose a hierarchy of dynamic medial axis structures at several spatio-temporal scales that can be modeled using a set of Linear Dynamical Systems (LDSs). We then propose novel discriminative metrics for comparing these sets of LDSs for the task of human activity recognition. Combined with simple classification frameworks, our proposed features and corresponding hierarchical dynamical models provide the highest human activity recognition rates as compared to state-of-the-art methods on several skeletal datasets.

AB - Over the last few years, with the immense popularity of the Kinect, there has been renewed interest in developing methods for human gesture and action recognition from 3D data. A number of approaches have been proposed that extract representative features from 3D depth data, a reconstructed 3D surface mesh or more commonly from the recovered estimate of the human skeleton. Recent advances in neuroscience have discovered a neural encoding of static 3D shapes in primate infero-temporal cortex that can be represented as a hierarchy of medial axis and surface features. We hypothesize a similar neural encoding might also exist for 3D shapes in motion and propose a hierarchy of dynamic medial axis structures at several spatio-temporal scales that can be modeled using a set of Linear Dynamical Systems (LDSs). We then propose novel discriminative metrics for comparing these sets of LDSs for the task of human activity recognition. Combined with simple classification frameworks, our proposed features and corresponding hierarchical dynamical models provide the highest human activity recognition rates as compared to state-of-the-art methods on several skeletal datasets.

UR - http://www.scopus.com/inward/record.url?scp=84884947071&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84884947071&partnerID=8YFLogxK

U2 - 10.1109/CVPRW.2013.153

DO - 10.1109/CVPRW.2013.153

M3 - Conference contribution

SN - 9780769549903

SP - 471

EP - 478

BT - IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops

ER -