Deep metric learning for video-based person re-identification

Naoki Kato, Kohei Hakozaki, Masamoto Tanabiki, Junko Furuyama, Yuji Sato, Yoshimitsu Aoki

Research output: Contribution to journalArticle

Abstract

This paper proposes a novel approach for video-based person re-identification that exploits deep convolutional neural networks to learn the similarity of persons observed from video camera. By Convolutional Neural Networks (CNN), each video sequence of a person is mapped to a Euclidean space where distances between feature embeddings directly correspond to measures of person similarity. By improved parameter learning method called Entire Triplet Loss, all possible triplets in the mini-batch are taken into account to update network parameters at once. This simple change of parameter updating method significantly improves network training, enabling the embeddings to be further discriminative. Experimental results show that proposed model achieves new state of the art identification rate on iLEDS-VID dataset and PRID-2011 dataset with 78.3%, 83.9% at rank 1, respectively.

Original languageEnglish
Pages (from-to)1117-1124
Number of pages8
JournalSeimitsu Kogaku Kaishi/Journal of the Japan Society for Precision Engineering
Volume83
Issue number12
DOIs
Publication statusPublished - 2017 Jan 1

Fingerprint

Neural networks
Video cameras
Identification (control systems)

Keywords

  • Computer vision
  • Convolutional neural networks
  • Machine learning
  • Metric learning
  • Pattern recognition
  • Person re-identification
  • Video recognition

ASJC Scopus subject areas

  • Mechanical Engineering

Cite this

Deep metric learning for video-based person re-identification. / Kato, Naoki; Hakozaki, Kohei; Tanabiki, Masamoto; Furuyama, Junko; Sato, Yuji; Aoki, Yoshimitsu.

In: Seimitsu Kogaku Kaishi/Journal of the Japan Society for Precision Engineering, Vol. 83, No. 12, 01.01.2017, p. 1117-1124.

Research output: Contribution to journalArticle

Kato, Naoki ; Hakozaki, Kohei ; Tanabiki, Masamoto ; Furuyama, Junko ; Sato, Yuji ; Aoki, Yoshimitsu. / Deep metric learning for video-based person re-identification. In: Seimitsu Kogaku Kaishi/Journal of the Japan Society for Precision Engineering. 2017 ; Vol. 83, No. 12. pp. 1117-1124.
@article{a6dc5e5f81324387ae10e613f7b059bb,
title = "Deep metric learning for video-based person re-identification",
abstract = "This paper proposes a novel approach for video-based person re-identification that exploits deep convolutional neural networks to learn the similarity of persons observed from video camera. By Convolutional Neural Networks (CNN), each video sequence of a person is mapped to a Euclidean space where distances between feature embeddings directly correspond to measures of person similarity. By improved parameter learning method called Entire Triplet Loss, all possible triplets in the mini-batch are taken into account to update network parameters at once. This simple change of parameter updating method significantly improves network training, enabling the embeddings to be further discriminative. Experimental results show that proposed model achieves new state of the art identification rate on iLEDS-VID dataset and PRID-2011 dataset with 78.3{\%}, 83.9{\%} at rank 1, respectively.",
keywords = "Computer vision, Convolutional neural networks, Machine learning, Metric learning, Pattern recognition, Person re-identification, Video recognition",
author = "Naoki Kato and Kohei Hakozaki and Masamoto Tanabiki and Junko Furuyama and Yuji Sato and Yoshimitsu Aoki",
year = "2017",
month = "1",
day = "1",
doi = "10.2493/jjspe.83.1117",
language = "English",
volume = "83",
pages = "1117--1124",
journal = "Seimitsu Kogaku Kaishi/Journal of the Japan Society for Precision Engineering",
issn = "0912-0289",
publisher = "Japan Society for Precision Engineering",
number = "12",

}

TY - JOUR

T1 - Deep metric learning for video-based person re-identification

AU - Kato, Naoki

AU - Hakozaki, Kohei

AU - Tanabiki, Masamoto

AU - Furuyama, Junko

AU - Sato, Yuji

AU - Aoki, Yoshimitsu

PY - 2017/1/1

Y1 - 2017/1/1

N2 - This paper proposes a novel approach for video-based person re-identification that exploits deep convolutional neural networks to learn the similarity of persons observed from video camera. By Convolutional Neural Networks (CNN), each video sequence of a person is mapped to a Euclidean space where distances between feature embeddings directly correspond to measures of person similarity. By improved parameter learning method called Entire Triplet Loss, all possible triplets in the mini-batch are taken into account to update network parameters at once. This simple change of parameter updating method significantly improves network training, enabling the embeddings to be further discriminative. Experimental results show that proposed model achieves new state of the art identification rate on iLEDS-VID dataset and PRID-2011 dataset with 78.3%, 83.9% at rank 1, respectively.

AB - This paper proposes a novel approach for video-based person re-identification that exploits deep convolutional neural networks to learn the similarity of persons observed from video camera. By Convolutional Neural Networks (CNN), each video sequence of a person is mapped to a Euclidean space where distances between feature embeddings directly correspond to measures of person similarity. By improved parameter learning method called Entire Triplet Loss, all possible triplets in the mini-batch are taken into account to update network parameters at once. This simple change of parameter updating method significantly improves network training, enabling the embeddings to be further discriminative. Experimental results show that proposed model achieves new state of the art identification rate on iLEDS-VID dataset and PRID-2011 dataset with 78.3%, 83.9% at rank 1, respectively.

KW - Computer vision

KW - Convolutional neural networks

KW - Machine learning

KW - Metric learning

KW - Pattern recognition

KW - Person re-identification

KW - Video recognition

UR - http://www.scopus.com/inward/record.url?scp=85037162957&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85037162957&partnerID=8YFLogxK

U2 - 10.2493/jjspe.83.1117

DO - 10.2493/jjspe.83.1117

M3 - Article

AN - SCOPUS:85037162957

VL - 83

SP - 1117

EP - 1124

JO - Seimitsu Kogaku Kaishi/Journal of the Japan Society for Precision Engineering

JF - Seimitsu Kogaku Kaishi/Journal of the Japan Society for Precision Engineering

SN - 0912-0289

IS - 12

ER -