Person re-identification based on deep multi-instance learning

Domonkos Varga, T. Szirányi

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

Abstract

Person re-identification is one of the widely studied research topic in the fields of computer vision and pattern recognition. In this paper, we present a deep multi-instance learning approach for person re-identification. Since most publicly available databases for pedestrian re-identification are not enough big, over-fitting problems occur in deep learning architectures. To tackle this problem, person re-identification is expressed as a deep multi-instance learning issue. Therefore, a multiscale feature learning process is introduced which is driven by optimizing a novel cost function. We report on experiments and comparisons to other state-of-the-art algorithms using publicly available databases such as VIPeR and ETHZ.

Original languageEnglish
Title of host publication25th European Signal Processing Conference, EUSIPCO 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1559-1563
Number of pages5
Volume2017-January
ISBN (Electronic)9780992862671
DOIs
Publication statusPublished - Oct 23 2017
Event25th European Signal Processing Conference, EUSIPCO 2017 - Kos, Greece
Duration: Aug 28 2017Sep 2 2017

Other

Other25th European Signal Processing Conference, EUSIPCO 2017
CountryGreece
CityKos
Period8/28/179/2/17

Fingerprint

Cost functions
Computer vision
Pattern recognition
Experiments
Deep learning

ASJC Scopus subject areas

  • Signal Processing

Cite this

Varga, D., & Szirányi, T. (2017). Person re-identification based on deep multi-instance learning. In 25th European Signal Processing Conference, EUSIPCO 2017 (Vol. 2017-January, pp. 1559-1563). [8081471] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.23919/EUSIPCO.2017.8081471

Person re-identification based on deep multi-instance learning. / Varga, Domonkos; Szirányi, T.

25th European Signal Processing Conference, EUSIPCO 2017. Vol. 2017-January Institute of Electrical and Electronics Engineers Inc., 2017. p. 1559-1563 8081471.

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

Varga, D & Szirányi, T 2017, Person re-identification based on deep multi-instance learning. in 25th European Signal Processing Conference, EUSIPCO 2017. vol. 2017-January, 8081471, Institute of Electrical and Electronics Engineers Inc., pp. 1559-1563, 25th European Signal Processing Conference, EUSIPCO 2017, Kos, Greece, 8/28/17. https://doi.org/10.23919/EUSIPCO.2017.8081471
Varga D, Szirányi T. Person re-identification based on deep multi-instance learning. In 25th European Signal Processing Conference, EUSIPCO 2017. Vol. 2017-January. Institute of Electrical and Electronics Engineers Inc. 2017. p. 1559-1563. 8081471 https://doi.org/10.23919/EUSIPCO.2017.8081471
Varga, Domonkos ; Szirányi, T. / Person re-identification based on deep multi-instance learning. 25th European Signal Processing Conference, EUSIPCO 2017. Vol. 2017-January Institute of Electrical and Electronics Engineers Inc., 2017. pp. 1559-1563
@inproceedings{8e7ec96f96b44241b4956adcf4a5fab6,
title = "Person re-identification based on deep multi-instance learning",
abstract = "Person re-identification is one of the widely studied research topic in the fields of computer vision and pattern recognition. In this paper, we present a deep multi-instance learning approach for person re-identification. Since most publicly available databases for pedestrian re-identification are not enough big, over-fitting problems occur in deep learning architectures. To tackle this problem, person re-identification is expressed as a deep multi-instance learning issue. Therefore, a multiscale feature learning process is introduced which is driven by optimizing a novel cost function. We report on experiments and comparisons to other state-of-the-art algorithms using publicly available databases such as VIPeR and ETHZ.",
author = "Domonkos Varga and T. Szir{\'a}nyi",
year = "2017",
month = "10",
day = "23",
doi = "10.23919/EUSIPCO.2017.8081471",
language = "English",
volume = "2017-January",
pages = "1559--1563",
booktitle = "25th European Signal Processing Conference, EUSIPCO 2017",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - Person re-identification based on deep multi-instance learning

AU - Varga, Domonkos

AU - Szirányi, T.

PY - 2017/10/23

Y1 - 2017/10/23

N2 - Person re-identification is one of the widely studied research topic in the fields of computer vision and pattern recognition. In this paper, we present a deep multi-instance learning approach for person re-identification. Since most publicly available databases for pedestrian re-identification are not enough big, over-fitting problems occur in deep learning architectures. To tackle this problem, person re-identification is expressed as a deep multi-instance learning issue. Therefore, a multiscale feature learning process is introduced which is driven by optimizing a novel cost function. We report on experiments and comparisons to other state-of-the-art algorithms using publicly available databases such as VIPeR and ETHZ.

AB - Person re-identification is one of the widely studied research topic in the fields of computer vision and pattern recognition. In this paper, we present a deep multi-instance learning approach for person re-identification. Since most publicly available databases for pedestrian re-identification are not enough big, over-fitting problems occur in deep learning architectures. To tackle this problem, person re-identification is expressed as a deep multi-instance learning issue. Therefore, a multiscale feature learning process is introduced which is driven by optimizing a novel cost function. We report on experiments and comparisons to other state-of-the-art algorithms using publicly available databases such as VIPeR and ETHZ.

UR - http://www.scopus.com/inward/record.url?scp=85041487640&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85041487640&partnerID=8YFLogxK

U2 - 10.23919/EUSIPCO.2017.8081471

DO - 10.23919/EUSIPCO.2017.8081471

M3 - Conference contribution

AN - SCOPUS:85041487640

VL - 2017-January

SP - 1559

EP - 1563

BT - 25th European Signal Processing Conference, EUSIPCO 2017

PB - Institute of Electrical and Electronics Engineers Inc.

ER -