End-to-end correspondence and relationship learning of mid-level deep features for person re-identification

Shan Lin; Chang-Tsun Li

doi:10.1109/DICTA.2017.8227426

End-to-end correspondence and relationship learning of mid-level deep features for person re-identification

Shan Lin, Chang-Tsun Li

University of Warwick

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

5 Citations (Scopus)

Abstract

In this paper, a unified deep convolutional architecture is proposed to address the problems in the person re-identification task. The proposed method adaptively learns the discriminative deep mid-level features of a person and constructs the correspondence features between an image pair in a data-driven manner. The previous Siamese structure deep learning approaches focus only on pair-wise matching between features. In our method, we consider the latent relationship between mid-level features and propose a network structure to automatically construct the correspondence features from all input features without a pre-defined matching function. The experimental results on three benchmarks VIPeR, CUHK01 and CUHK03 show that our unified approach improves over the previous state-of-the-art methods.

Original language	English
Title of host publication	Proceedings of the 2017 International Conference on digital image computing
Subtitle of host publication	Techniques and applications
Place of Publication	United States
Publisher	IEEE, Institute of Electrical and Electronics Engineers
Pages	1-6
Number of pages	6
ISBN (Electronic)	9781538628393
ISBN (Print)	9781538628409
DOIs	https://doi.org/10.1109/DICTA.2017.8227426
Publication status	Published - 21 Dec 2017
Event	2017 International Conference on Digital Image Computing: Techniques and Applications (DICTA) - Novotel Sydney Manly Pacific, Sydney, Australia Duration: 29 Nov 2017 → 01 Dec 2017 http://dicta2017.dictaconference.org/index.html (Conference website) http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=8226656 (Conference proceedings)

Conference

Conference	2017 International Conference on Digital Image Computing: Techniques and Applications (DICTA)
Country/Territory	Australia
City	Sydney
Period	29/11/17 → 01/12/17
Other	The International Conference on Digital Image Computing: Techniques and Applications (DICTA) is the main Australian Conference on computer vision, image processing, pattern recognition, and related areas. DICTA was established in 1991 as the premier conference of the Australian Pattern Recognition Society (APRS).
Internet address	http://dicta2017.dictaconference.org/index.html (Conference website) http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=8226656 (Conference proceedings)

Access to Document

10.1109/DICTA.2017.8227426

Cite this

@inproceedings{2b475a74ce75460485534ad6cfc91886,

title = "End-to-end correspondence and relationship learning of mid-level deep features for person re-identification",

abstract = "In this paper, a unified deep convolutional architecture is proposed to address the problems in the person re-identification task. The proposed method adaptively learns the discriminative deep mid-level features of a person and constructs the correspondence features between an image pair in a data-driven manner. The previous Siamese structure deep learning approaches focus only on pair-wise matching between features. In our method, we consider the latent relationship between mid-level features and propose a network structure to automatically construct the correspondence features from all input features without a pre-defined matching function. The experimental results on three benchmarks VIPeR, CUHK01 and CUHK03 show that our unified approach improves over the previous state-of-the-art methods.",

keywords = "Feature extraction, Cameras, Training, Measurement, Computer architecture, Shape",

author = "Shan Lin and Chang-Tsun Li",

note = "Includes bibliographical references.; 2017 International Conference on Digital Image Computing: Techniques and Applications (DICTA) ; Conference date: 29-11-2017 Through 01-12-2017",

year = "2017",

month = dec,

day = "21",

doi = "10.1109/DICTA.2017.8227426",

language = "English",

isbn = "9781538628409",

pages = "1--6",

booktitle = "Proceedings of the 2017 International Conference on digital image computing",

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

address = "United States",

url = "http://dicta2017.dictaconference.org/index.html, http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=8226656",

}

Lin, S & Li, C-T 2017, End-to-end correspondence and relationship learning of mid-level deep features for person re-identification. in Proceedings of the 2017 International Conference on digital image computing: Techniques and applications. IEEE, Institute of Electrical and Electronics Engineers, United States, pp. 1-6, 2017 International Conference on Digital Image Computing: Techniques and Applications (DICTA), Sydney, New South Wales, Australia, 29/11/17. https://doi.org/10.1109/DICTA.2017.8227426

End-to-end correspondence and relationship learning of mid-level deep features for person re-identification. / Lin, Shan; Li, Chang-Tsun.
Proceedings of the 2017 International Conference on digital image computing: Techniques and applications. United States: IEEE, Institute of Electrical and Electronics Engineers, 2017. p. 1-6.

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

TY - GEN

T1 - End-to-end correspondence and relationship learning of mid-level deep features for person re-identification

AU - Lin, Shan

AU - Li, Chang-Tsun

N1 - Includes bibliographical references.

PY - 2017/12/21

Y1 - 2017/12/21

N2 - In this paper, a unified deep convolutional architecture is proposed to address the problems in the person re-identification task. The proposed method adaptively learns the discriminative deep mid-level features of a person and constructs the correspondence features between an image pair in a data-driven manner. The previous Siamese structure deep learning approaches focus only on pair-wise matching between features. In our method, we consider the latent relationship between mid-level features and propose a network structure to automatically construct the correspondence features from all input features without a pre-defined matching function. The experimental results on three benchmarks VIPeR, CUHK01 and CUHK03 show that our unified approach improves over the previous state-of-the-art methods.

AB - In this paper, a unified deep convolutional architecture is proposed to address the problems in the person re-identification task. The proposed method adaptively learns the discriminative deep mid-level features of a person and constructs the correspondence features between an image pair in a data-driven manner. The previous Siamese structure deep learning approaches focus only on pair-wise matching between features. In our method, we consider the latent relationship between mid-level features and propose a network structure to automatically construct the correspondence features from all input features without a pre-defined matching function. The experimental results on three benchmarks VIPeR, CUHK01 and CUHK03 show that our unified approach improves over the previous state-of-the-art methods.

KW - Feature extraction

KW - Cameras

KW - Training

KW - Measurement

KW - Computer architecture

KW - Shape

UR - http://dicta2017.dictaconference.org/fullprogram.html

UR - http://dicta2017.dictaconference.org/camerasubmission.html

UR - http://dicta2017.dictaconference.org/callfor.html

U2 - 10.1109/DICTA.2017.8227426

DO - 10.1109/DICTA.2017.8227426

M3 - Conference paper

SN - 9781538628409

SP - 1

EP - 6

BT - Proceedings of the 2017 International Conference on digital image computing

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - United States

T2 - 2017 International Conference on Digital Image Computing: Techniques and Applications (DICTA)

Y2 - 29 November 2017 through 1 December 2017

ER -

End-to-end correspondence and relationship learning of mid-level deep features for person re-identification

Abstract

Conference

Access to Document

Other files and links

Fingerprint

Cite this