Learning optimised representations for view-invariant gait recognition

Ning Jia; Victor Sanchez; Chang Tsun Li

doi:10.1109/BTAS.2017.8272769

Learning optimised representations for view-invariant gait recognition

Ning Jia, Victor Sanchez, Chang Tsun Li

University of Warwick

Research output: Book chapter/Published conference paper › Conference paper › peer-review

10 Citations (Scopus)

Abstract

Gait recognition can be performed without subject cooperation under harsh conditions, thus it is an important tool in forensic gait analysis, security control, and other commercial applications. One critical issue that prevents gait recognition systems from being widely accepted is the performance drop when the camera viewpoint varies between the registered templates and the query data. In this paper, we explore the potential of combining feature optimisers and representations learned by convolutional neural networks (CNN) to achieve efficient view-invariant gait recognition. The experimental results indicate that CNN learns highly discriminative representations across moderate view variations, and these representations can be further improved using view-invariant feature selectors, achieving a high matching accuracy across views.

Original language	English
Title of host publication	2017 International Joint Conference on Biometrics (IJCB)
Place of Publication	United States
Publisher	IEEE, Institute of Electrical and Electronics Engineers
Pages	774-780
Number of pages	7
ISBN (Electronic)	9781538611241
ISBN (Print)	9781538611258 (Print on demand)
DOIs	https://doi.org/10.1109/BTAS.2017.8272769
Publication status	Published - 01 Feb 2018
Event	IEEE International Joint Conference on Biometrics: IJCB 2017 - Denver Marriott Tech Center , Denver, United States Duration: 01 Oct 2017 → 04 Oct 2017 http://www.ijcb2017.org/ijcb2017/index.php (Conference website) https://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=8263072 (Conference proceedings)

Conference

Conference	IEEE International Joint Conference on Biometrics
Country/Territory	United States
City	Denver
Period	01/10/17 → 04/10/17
Other	The International Joint Conference on Biometrics (IJCB 2017) combines two major biometrics research annual conferences, the Biometrics Theory, Applications and Systems (BTAS) conference and the International Conference on Biometrics (ICB). The blending of these two conferences in 2017 is through special agreement between the IEEE Biometrics Council and the IAPR TC-4, and presents an exciting event for the entire worldwide biometrics research community. This is the third such joint conference, following IJCB 2014 in Clearwater Florida and the first IJCB 2011 in Washington DC. This conference is a result of major worldwide consensus to join the two major biometrics meetings and to establish IJCB as a venue for presenting biometrics research results of highest quality.
Internet address	http://www.ijcb2017.org/ijcb2017/index.php (Conference website) https://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=8263072 (Conference proceedings)

Access to Document

10.1109/BTAS.2017.8272769

Cite this

@inproceedings{6942b855ddff4e99b5b7bef727c9e1ed,

title = "Learning optimised representations for view-invariant gait recognition",

abstract = "Gait recognition can be performed without subject cooperation under harsh conditions, thus it is an important tool in forensic gait analysis, security control, and other commercial applications. One critical issue that prevents gait recognition systems from being widely accepted is the performance drop when the camera viewpoint varies between the registered templates and the query data. In this paper, we explore the potential of combining feature optimisers and representations learned by convolutional neural networks (CNN) to achieve efficient view-invariant gait recognition. The experimental results indicate that CNN learns highly discriminative representations across moderate view variations, and these representations can be further improved using view-invariant feature selectors, achieving a high matching accuracy across views.",

keywords = "Gait recognition, Probes, Cameras, Training, Computational modeling, Solid modeling",

author = "Ning Jia and Victor Sanchez and Li, {Chang Tsun}",

note = "Includes bibliographical references.; IEEE International Joint Conference on Biometrics : IJCB 2017 ; Conference date: 01-10-2017 Through 04-10-2017",

year = "2018",

month = feb,

day = "1",

doi = "10.1109/BTAS.2017.8272769",

language = "English",

isbn = "9781538611258 (Print on demand)",

pages = "774--780",

booktitle = "2017 International Joint Conference on Biometrics (IJCB)",

publisher = "IEEE, Institute of Electrical and Electronics Engineers",

address = "United States",

url = "http://www.ijcb2017.org/ijcb2017/index.php, https://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=8263072",

}

Jia, N, Sanchez, V & Li, CT 2018, Learning optimised representations for view-invariant gait recognition. in 2017 International Joint Conference on Biometrics (IJCB)., 8272769, IEEE, Institute of Electrical and Electronics Engineers, United States, pp. 774-780, IEEE International Joint Conference on Biometrics, Denver, Colorado, United States, 01/10/17. https://doi.org/10.1109/BTAS.2017.8272769

TY - GEN

T1 - Learning optimised representations for view-invariant gait recognition

AU - Jia, Ning

AU - Sanchez, Victor

AU - Li, Chang Tsun

N1 - Includes bibliographical references.

PY - 2018/2/1

Y1 - 2018/2/1

N2 - Gait recognition can be performed without subject cooperation under harsh conditions, thus it is an important tool in forensic gait analysis, security control, and other commercial applications. One critical issue that prevents gait recognition systems from being widely accepted is the performance drop when the camera viewpoint varies between the registered templates and the query data. In this paper, we explore the potential of combining feature optimisers and representations learned by convolutional neural networks (CNN) to achieve efficient view-invariant gait recognition. The experimental results indicate that CNN learns highly discriminative representations across moderate view variations, and these representations can be further improved using view-invariant feature selectors, achieving a high matching accuracy across views.

AB - Gait recognition can be performed without subject cooperation under harsh conditions, thus it is an important tool in forensic gait analysis, security control, and other commercial applications. One critical issue that prevents gait recognition systems from being widely accepted is the performance drop when the camera viewpoint varies between the registered templates and the query data. In this paper, we explore the potential of combining feature optimisers and representations learned by convolutional neural networks (CNN) to achieve efficient view-invariant gait recognition. The experimental results indicate that CNN learns highly discriminative representations across moderate view variations, and these representations can be further improved using view-invariant feature selectors, achieving a high matching accuracy across views.

KW - Gait recognition

KW - Probes

KW - Cameras

KW - Training

KW - Computational modeling

KW - Solid modeling

UR - http://www.scopus.com/inward/record.url?scp=85046289282&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85046289282&partnerID=8YFLogxK

U2 - 10.1109/BTAS.2017.8272769

DO - 10.1109/BTAS.2017.8272769

M3 - Conference paper

AN - SCOPUS:85046289282

SN - 9781538611258 (Print on demand)

SP - 774

EP - 780

BT - 2017 International Joint Conference on Biometrics (IJCB)

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - United States

T2 - IEEE International Joint Conference on Biometrics

Y2 - 1 October 2017 through 4 October 2017

ER -

Learning optimised representations for view-invariant gait recognition

Abstract

Conference

Access to Document

Other files and links

Fingerprint

Cite this