Task-specific neural networks for pose estimation in person re-identification task

Kai Lv; Hao Sheng; Yanwei Zheng; Zhang Xiong; Wei Li; Wei Ke

doi:10.1007/978-3-319-77380-3_76

Task-specific neural networks for pose estimation in person re-identification task

Kai Lv, Hao Sheng, Yanwei Zheng, Zhang Xiong, Wei Li, Wei Ke

Faculty of Applied Sciences

Beihang University

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

Abstract

Person re-identification is a challenging task because of severe appearance changes of a person due to diverse camera viewpoints and person poses. To alleviate the impact of different poses, more and more studies have been done in pose estimation. In this work, we present three task-specific neural networks (TNN) algorithm designed to address the problem of pose estimation for re-identification in both single-shot and multi-shot matching. In order to recognize the human pose as one of the four classes (front, back, left, right), a PoseNet-A is first required to estimate the pose as class front-back or class left-right. Based on the results, we select the appropriate network (PoseNet-B1, PoseNet-B2) to obtain the final pose. According to the results, our method achieves very good results on a large data set (CUHK03-Pose). One thing that needs to be pointed out is that we build the dataset CUHK03-Pose which is based on the person re-identification dataset CUHK03.

Original language	English
Title of host publication	Advances in Multimedia Information Processing – PCM 2017 - 18th Pacific-Rim Conference on Multimedia, Revised Selected Papers
Editors	Bing Zeng, Hongliang Li, Abdulmotaleb El Saddik, Xiaopeng Fan, Shuqiang Jiang, Qingming Huang
Publisher	Springer Verlag
Pages	792-801
Number of pages	10
ISBN (Print)	9783319773797
DOIs	https://doi.org/10.1007/978-3-319-77380-3_76
Publication status	Published - 2018
Event	18th Pacific-Rim Conference on Multimedia, PCM 2017 - Harbin, China Duration: 28 Sept 2017 → 29 Sept 2017

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10735 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	18th Pacific-Rim Conference on Multimedia, PCM 2017
Country/Territory	China
City	Harbin
Period	28/09/17 → 29/09/17

Keywords

Deep learning
Pose estimation
Re-identification
Single-shot

Access to Document

10.1007/978-3-319-77380-3_76

Cite this

Lv, K., Sheng, H., Zheng, Y., Xiong, Z., Li, W., & Ke, W. (2018). Task-specific neural networks for pose estimation in person re-identification task. In B. Zeng, H. Li, A. El Saddik, X. Fan, S. Jiang, & Q. Huang (Eds.), Advances in Multimedia Information Processing – PCM 2017 - 18th Pacific-Rim Conference on Multimedia, Revised Selected Papers (pp. 792-801). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10735 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-77380-3_76

Lv, Kai ; Sheng, Hao ; Zheng, Yanwei et al. / Task-specific neural networks for pose estimation in person re-identification task. Advances in Multimedia Information Processing – PCM 2017 - 18th Pacific-Rim Conference on Multimedia, Revised Selected Papers. editor / Bing Zeng ; Hongliang Li ; Abdulmotaleb El Saddik ; Xiaopeng Fan ; Shuqiang Jiang ; Qingming Huang. Springer Verlag, 2018. pp. 792-801 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{78037dddd4504caf9e0aeb4b8d30e358,

title = "Task-specific neural networks for pose estimation in person re-identification task",

abstract = "Person re-identification is a challenging task because of severe appearance changes of a person due to diverse camera viewpoints and person poses. To alleviate the impact of different poses, more and more studies have been done in pose estimation. In this work, we present three task-specific neural networks (TNN) algorithm designed to address the problem of pose estimation for re-identification in both single-shot and multi-shot matching. In order to recognize the human pose as one of the four classes (front, back, left, right), a PoseNet-A is first required to estimate the pose as class front-back or class left-right. Based on the results, we select the appropriate network (PoseNet-B1, PoseNet-B2) to obtain the final pose. According to the results, our method achieves very good results on a large data set (CUHK03-Pose). One thing that needs to be pointed out is that we build the dataset CUHK03-Pose which is based on the person re-identification dataset CUHK03.",

keywords = "Deep learning, Pose estimation, Re-identification, Single-shot",

author = "Kai Lv and Hao Sheng and Yanwei Zheng and Zhang Xiong and Wei Li and Wei Ke",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing AG, part of Springer Nature 2018.; 18th Pacific-Rim Conference on Multimedia, PCM 2017 ; Conference date: 28-09-2017 Through 29-09-2017",

year = "2018",

doi = "10.1007/978-3-319-77380-3_76",

language = "English",

isbn = "9783319773797",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "792--801",

editor = "Bing Zeng and Hongliang Li and {El Saddik}, Abdulmotaleb and Xiaopeng Fan and Shuqiang Jiang and Qingming Huang",

booktitle = "Advances in Multimedia Information Processing – PCM 2017 - 18th Pacific-Rim Conference on Multimedia, Revised Selected Papers",

address = "Germany",

}

Lv, K, Sheng, H, Zheng, Y, Xiong, Z, Li, W & Ke, W 2018, Task-specific neural networks for pose estimation in person re-identification task. in B Zeng, H Li, A El Saddik, X Fan, S Jiang & Q Huang (eds), Advances in Multimedia Information Processing – PCM 2017 - 18th Pacific-Rim Conference on Multimedia, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10735 LNCS, Springer Verlag, pp. 792-801, 18th Pacific-Rim Conference on Multimedia, PCM 2017, Harbin, China, 28/09/17. https://doi.org/10.1007/978-3-319-77380-3_76

Task-specific neural networks for pose estimation in person re-identification task. / Lv, Kai; Sheng, Hao; Zheng, Yanwei et al.
Advances in Multimedia Information Processing – PCM 2017 - 18th Pacific-Rim Conference on Multimedia, Revised Selected Papers. ed. / Bing Zeng; Hongliang Li; Abdulmotaleb El Saddik; Xiaopeng Fan; Shuqiang Jiang; Qingming Huang. Springer Verlag, 2018. p. 792-801 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10735 LNCS).

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

TY - GEN

T1 - Task-specific neural networks for pose estimation in person re-identification task

AU - Lv, Kai

AU - Sheng, Hao

AU - Zheng, Yanwei

AU - Xiong, Zhang

AU - Li, Wei

AU - Ke, Wei

N1 - Publisher Copyright: © Springer International Publishing AG, part of Springer Nature 2018.

PY - 2018

Y1 - 2018

N2 - Person re-identification is a challenging task because of severe appearance changes of a person due to diverse camera viewpoints and person poses. To alleviate the impact of different poses, more and more studies have been done in pose estimation. In this work, we present three task-specific neural networks (TNN) algorithm designed to address the problem of pose estimation for re-identification in both single-shot and multi-shot matching. In order to recognize the human pose as one of the four classes (front, back, left, right), a PoseNet-A is first required to estimate the pose as class front-back or class left-right. Based on the results, we select the appropriate network (PoseNet-B1, PoseNet-B2) to obtain the final pose. According to the results, our method achieves very good results on a large data set (CUHK03-Pose). One thing that needs to be pointed out is that we build the dataset CUHK03-Pose which is based on the person re-identification dataset CUHK03.

AB - Person re-identification is a challenging task because of severe appearance changes of a person due to diverse camera viewpoints and person poses. To alleviate the impact of different poses, more and more studies have been done in pose estimation. In this work, we present three task-specific neural networks (TNN) algorithm designed to address the problem of pose estimation for re-identification in both single-shot and multi-shot matching. In order to recognize the human pose as one of the four classes (front, back, left, right), a PoseNet-A is first required to estimate the pose as class front-back or class left-right. Based on the results, we select the appropriate network (PoseNet-B1, PoseNet-B2) to obtain the final pose. According to the results, our method achieves very good results on a large data set (CUHK03-Pose). One thing that needs to be pointed out is that we build the dataset CUHK03-Pose which is based on the person re-identification dataset CUHK03.

KW - Deep learning

KW - Pose estimation

KW - Re-identification

KW - Single-shot

UR - http://www.scopus.com/inward/record.url?scp=85047484475&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-77380-3_76

DO - 10.1007/978-3-319-77380-3_76

M3 - Conference contribution

AN - SCOPUS:85047484475

SN - 9783319773797

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 792

EP - 801

BT - Advances in Multimedia Information Processing – PCM 2017 - 18th Pacific-Rim Conference on Multimedia, Revised Selected Papers

A2 - Zeng, Bing

A2 - Li, Hongliang

A2 - El Saddik, Abdulmotaleb

A2 - Fan, Xiaopeng

A2 - Jiang, Shuqiang

A2 - Huang, Qingming

PB - Springer Verlag

T2 - 18th Pacific-Rim Conference on Multimedia, PCM 2017

Y2 - 28 September 2017 through 29 September 2017

ER -

Lv K, Sheng H, Zheng Y, Xiong Z, Li W, Ke W. Task-specific neural networks for pose estimation in person re-identification task. In Zeng B, Li H, El Saddik A, Fan X, Jiang S, Huang Q, editors, Advances in Multimedia Information Processing – PCM 2017 - 18th Pacific-Rim Conference on Multimedia, Revised Selected Papers. Springer Verlag. 2018. p. 792-801. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-77380-3_76

Task-specific neural networks for pose estimation in person re-identification task

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this