An Approach for Gait Anonymization Using Deep Learning

Ngoc-Dung  T. Tieu; Huy H. Nguyen; Hoang-Quoc  Nguyen-Son; Junichi Yamagishi; Isao Echizen

doi:10.1109/WIFS.2017.8267657

An Approach for Gait Anonymization Using Deep Learning

Ngoc-Dung T. Tieu, Huy H. Nguyen, Hoang-Quoc Nguyen-Son, Junichi Yamagishi, Isao Echizen

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract / Description of output

The human gait has become another biometric trait used in security systems because it is unique to each person and can be recognized at a distance. However, a bad actor could use a gait recognition system to identify a person on the basis of his or her gait. We have developed a gait anonymization method that prevents unauthorized gait recognition. It modifies the gait so that the person cannot be identified while maintaining the naturalness of the gait. The modification is done by adding another gait, called “noise gait”. A convolutional neural network makes this modification by taking two gaits as input, the original gait and the noise gait, and outputting an anonymized gait. The proposed method was evaluated using the success rate and mean opinion score (MOS). The success rate is the rate of failed gait recognition, and the MOS is a measure of the naturalness of the anonymized gait. In our experiments, the success rate achieved 98.86% at most while the highest naturalness score is 3.73 in the MOS scale. These findings should open new research directions regarding privacy protection related to gait recognition.
Index Terms—gait; biometric trait; security; gait anonymization; deep learning

Original language	English
Title of host publication	9th IEEE International Workshop on Information Forensics and Security (WIFS) 2017
Place of Publication	Rennes, France
Publisher	Institute of Electrical and Electronics Engineers
Number of pages	6
ISBN (Electronic)	978-1-5090-6769-5
ISBN (Print)	978-1-5090-6770-1
DOIs	https://doi.org/10.1109/WIFS.2017.8267657
Publication status	Published - 25 Jan 2018
Event	9th IEEE International Workshop on Information Forensics and Security - , Hong Kong Duration: 4 Dec 2017 → 7 Dec 2017 https://project.inria.fr/wifs2017/

Publication series

Name
Publisher	IEEE
ISSN (Electronic)	2157-4774

Conference

Conference	9th IEEE International Workshop on Information Forensics and Security
Abbreviated title	WIFS 2017
Country/Territory	Hong Kong
Period	4/12/17 → 7/12/17
Internet address	https://project.inria.fr/wifs2017/

Access to Document

10.1109/WIFS.2017.8267657

An Approach for Gait AnonymizationAccepted author manuscript, 1.26 MB

https://ieeexplore.ieee.org/document/8267657/

Cite this

@inproceedings{4bd08405da3340e4b57a24e571461e97,

title = "An Approach for Gait Anonymization Using Deep Learning",

abstract = "The human gait has become another biometric trait used in security systems because it is unique to each person and can be recognized at a distance. However, a bad actor could use a gait recognition system to identify a person on the basis of his or her gait. We have developed a gait anonymization method that prevents unauthorized gait recognition. It modifies the gait so that the person cannot be identified while maintaining the naturalness of the gait. The modification is done by adding another gait, called “noise gait”. A convolutional neural network makes this modification by taking two gaits as input, the original gait and the noise gait, and outputting an anonymized gait. The proposed method was evaluated using the success rate and mean opinion score (MOS). The success rate is the rate of failed gait recognition, and the MOS is a measure of the naturalness of the anonymized gait. In our experiments, the success rate achieved 98.86% at most while the highest naturalness score is 3.73 in the MOS scale. These findings should open new research directions regarding privacy protection related to gait recognition.Index Terms—gait; biometric trait; security; gait anonymization; deep learning",

author = "{T. Tieu}, Ngoc-Dung and {H. Nguyen}, Huy and Hoang-Quoc Nguyen-Son and Junichi Yamagishi and Isao Echizen",

year = "2018",

month = jan,

day = "25",

doi = "10.1109/WIFS.2017.8267657",

language = "English",

isbn = "978-1-5090-6770-1",

publisher = "Institute of Electrical and Electronics Engineers",

booktitle = "9th IEEE International Workshop on Information Forensics and Security (WIFS) 2017",

address = "United States",

note = "9th IEEE International Workshop on Information Forensics and Security , WIFS 2017 ; Conference date: 04-12-2017 Through 07-12-2017",

url = "https://project.inria.fr/wifs2017/",

}

T. Tieu, N-D, H. Nguyen, H, Nguyen-Son, H-Q, Yamagishi, J & Echizen, I 2018, An Approach for Gait Anonymization Using Deep Learning. in 9th IEEE International Workshop on Information Forensics and Security (WIFS) 2017. Institute of Electrical and Electronics Engineers, Rennes, France, 9th IEEE International Workshop on Information Forensics and Security , Hong Kong, 4/12/17. https://doi.org/10.1109/WIFS.2017.8267657

TY - GEN

T1 - An Approach for Gait Anonymization Using Deep Learning

AU - T. Tieu, Ngoc-Dung

AU - H. Nguyen, Huy

AU - Nguyen-Son, Hoang-Quoc

AU - Yamagishi, Junichi

AU - Echizen, Isao

PY - 2018/1/25

Y1 - 2018/1/25

N2 - The human gait has become another biometric trait used in security systems because it is unique to each person and can be recognized at a distance. However, a bad actor could use a gait recognition system to identify a person on the basis of his or her gait. We have developed a gait anonymization method that prevents unauthorized gait recognition. It modifies the gait so that the person cannot be identified while maintaining the naturalness of the gait. The modification is done by adding another gait, called “noise gait”. A convolutional neural network makes this modification by taking two gaits as input, the original gait and the noise gait, and outputting an anonymized gait. The proposed method was evaluated using the success rate and mean opinion score (MOS). The success rate is the rate of failed gait recognition, and the MOS is a measure of the naturalness of the anonymized gait. In our experiments, the success rate achieved 98.86% at most while the highest naturalness score is 3.73 in the MOS scale. These findings should open new research directions regarding privacy protection related to gait recognition.Index Terms—gait; biometric trait; security; gait anonymization; deep learning

AB - The human gait has become another biometric trait used in security systems because it is unique to each person and can be recognized at a distance. However, a bad actor could use a gait recognition system to identify a person on the basis of his or her gait. We have developed a gait anonymization method that prevents unauthorized gait recognition. It modifies the gait so that the person cannot be identified while maintaining the naturalness of the gait. The modification is done by adding another gait, called “noise gait”. A convolutional neural network makes this modification by taking two gaits as input, the original gait and the noise gait, and outputting an anonymized gait. The proposed method was evaluated using the success rate and mean opinion score (MOS). The success rate is the rate of failed gait recognition, and the MOS is a measure of the naturalness of the anonymized gait. In our experiments, the success rate achieved 98.86% at most while the highest naturalness score is 3.73 in the MOS scale. These findings should open new research directions regarding privacy protection related to gait recognition.Index Terms—gait; biometric trait; security; gait anonymization; deep learning

U2 - 10.1109/WIFS.2017.8267657

DO - 10.1109/WIFS.2017.8267657

M3 - Conference contribution

SN - 978-1-5090-6770-1

BT - 9th IEEE International Workshop on Information Forensics and Security (WIFS) 2017

PB - Institute of Electrical and Electronics Engineers

CY - Rennes, France

T2 - 9th IEEE International Workshop on Information Forensics and Security

Y2 - 4 December 2017 through 7 December 2017

ER -

An Approach for Gait Anonymization Using Deep Learning

Abstract / Description of output

Publication series

Conference

Access to Document

Fingerprint

Cite this