Deep metric learning with online hard and soft selection for person re-identification

Mingyang Yu, Seiichiro Kamata

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

Abstract

Deep metric learning has been widely used for image retrieval and verification tasks. Traditional contrastive loss and triplet loss depend highly on the selection of pair/triplet images. It makes the training process unstable and uncomplete. In this paper, we propose a novel global level loss function that considers histograms for intra distances within class and inter distances between different classes. We compared two forms of global level loss (hard selection based loss and soft selection based loss) and both achieved better result than traditional triplet loss, multi-class N pair loss and other related works. The experiment is conducted on the person re-identification dataset Market 1501 and DukeMTMC-reID.

Original languageEnglish
Title of host publication2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages426-431
Number of pages6
ISBN (Electronic)9781538651612
DOIs
Publication statusPublished - 2019 Feb 12
EventJoint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018 - Kitakyushu, Japan
Duration: 2018 Jun 252018 Jun 28

Publication series

Name2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018

Conference

ConferenceJoint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018
CountryJapan
CityKitakyushu
Period18/6/2518/6/28

Fingerprint

Person
Metric
Image Retrieval
Multi-class
Loss Function
Learning
Image retrieval
Histogram
Unstable
Experiment
Experiments
Class

Keywords

  • Deep metric learning
  • Feature embedding
  • Per-son Re-identification

ASJC Scopus subject areas

  • Signal Processing
  • Control and Optimization
  • Artificial Intelligence
  • Computer Vision and Pattern Recognition
  • Information Systems

Cite this

Yu, M., & Kamata, S. (2019). Deep metric learning with online hard and soft selection for person re-identification. In 2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018 (pp. 426-431). [8641037] (2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICIEV.2018.8641037

Deep metric learning with online hard and soft selection for person re-identification. / Yu, Mingyang; Kamata, Seiichiro.

2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018. Institute of Electrical and Electronics Engineers Inc., 2019. p. 426-431 8641037 (2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018).

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

Yu, M & Kamata, S 2019, Deep metric learning with online hard and soft selection for person re-identification. in 2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018., 8641037, 2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018, Institute of Electrical and Electronics Engineers Inc., pp. 426-431, Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018, Kitakyushu, Japan, 18/6/25. https://doi.org/10.1109/ICIEV.2018.8641037
Yu M, Kamata S. Deep metric learning with online hard and soft selection for person re-identification. In 2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018. Institute of Electrical and Electronics Engineers Inc. 2019. p. 426-431. 8641037. (2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018). https://doi.org/10.1109/ICIEV.2018.8641037
Yu, Mingyang ; Kamata, Seiichiro. / Deep metric learning with online hard and soft selection for person re-identification. 2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 426-431 (2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018).
@inproceedings{569051b246ed4806af4d6ed7dbd9d88a,
title = "Deep metric learning with online hard and soft selection for person re-identification",
abstract = "Deep metric learning has been widely used for image retrieval and verification tasks. Traditional contrastive loss and triplet loss depend highly on the selection of pair/triplet images. It makes the training process unstable and uncomplete. In this paper, we propose a novel global level loss function that considers histograms for intra distances within class and inter distances between different classes. We compared two forms of global level loss (hard selection based loss and soft selection based loss) and both achieved better result than traditional triplet loss, multi-class N pair loss and other related works. The experiment is conducted on the person re-identification dataset Market 1501 and DukeMTMC-reID.",
keywords = "Deep metric learning, Feature embedding, Per-son Re-identification",
author = "Mingyang Yu and Seiichiro Kamata",
year = "2019",
month = "2",
day = "12",
doi = "10.1109/ICIEV.2018.8641037",
language = "English",
series = "2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "426--431",
booktitle = "2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018",

}

TY - GEN

T1 - Deep metric learning with online hard and soft selection for person re-identification

AU - Yu, Mingyang

AU - Kamata, Seiichiro

PY - 2019/2/12

Y1 - 2019/2/12

N2 - Deep metric learning has been widely used for image retrieval and verification tasks. Traditional contrastive loss and triplet loss depend highly on the selection of pair/triplet images. It makes the training process unstable and uncomplete. In this paper, we propose a novel global level loss function that considers histograms for intra distances within class and inter distances between different classes. We compared two forms of global level loss (hard selection based loss and soft selection based loss) and both achieved better result than traditional triplet loss, multi-class N pair loss and other related works. The experiment is conducted on the person re-identification dataset Market 1501 and DukeMTMC-reID.

AB - Deep metric learning has been widely used for image retrieval and verification tasks. Traditional contrastive loss and triplet loss depend highly on the selection of pair/triplet images. It makes the training process unstable and uncomplete. In this paper, we propose a novel global level loss function that considers histograms for intra distances within class and inter distances between different classes. We compared two forms of global level loss (hard selection based loss and soft selection based loss) and both achieved better result than traditional triplet loss, multi-class N pair loss and other related works. The experiment is conducted on the person re-identification dataset Market 1501 and DukeMTMC-reID.

KW - Deep metric learning

KW - Feature embedding

KW - Per-son Re-identification

UR - http://www.scopus.com/inward/record.url?scp=85063221482&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85063221482&partnerID=8YFLogxK

U2 - 10.1109/ICIEV.2018.8641037

DO - 10.1109/ICIEV.2018.8641037

M3 - Conference contribution

T3 - 2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018

SP - 426

EP - 431

BT - 2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018

PB - Institute of Electrical and Electronics Engineers Inc.

ER -