Multiple Object Tracking by Joint Head, Body Detection and Re-Identification

Zuode Liu; Honghai Liu; Weihong Ren; Hui Chang; Yuhang Shi; Ruihan Lin; Wenhao Wu

doi:10.1007/978-3-031-13841-6_16

Multiple Object Tracking by Joint Head, Body Detection and Re-Identification

Zuode Liu
, Honghai Liu^*
, Weihong Ren
, Hui Chang
, Yuhang Shi
, Ruihan Lin
, Wenhao Wu

^*Corresponding author for this work

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

1 Scopus citations

Abstract

Multi-object tracking (MOT) is an important problem in computer vision which has a wide range of applications. Formulating MOT as multi-task learning of object detection and re-Identification (re-ID) in a single network is appealing since it achieves real-time but effective inference on detection and tracking. However, in crowd scenes, the existing MOT methods usually fail to locate occluded objects, which also results in bad effects on the re-ID task. To solve people tracking in crowd scenes, we present a model called HBR (Head-Body-ReID Joint Tracking) to jointly formulates head detection, body detection and re-ID tasks into an uniform framework. Human heads are hardly affected by occlusions in crowd scenes, and they can provide informative clues for whole body detection. The experimental results on MOT17 and MOT20 show that our proposed model performs better than the state-of-the-arts.

Original language	English
Title of host publication	Intelligent Robotics and Applications - 15th International Conference, ICIRA 2022, Proceedings
Editors	Honghai Liu, Weihong Ren, Zhouping Yin, Lianqing Liu, Li Jiang, Guoying Gu, Xinyu Wu
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	171-180
Number of pages	10
ISBN (Print)	9783031138409
DOIs	https://doi.org/10.1007/978-3-031-13841-6_16
State	Published - 2022
Externally published	Yes
Event	15th International Conference on Intelligent Robotics and Applications, ICIRA 2022 - Harbin, China Duration: 1 Aug 2022 → 3 Aug 2022

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13458 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	15th International Conference on Intelligent Robotics and Applications, ICIRA 2022
Country/Territory	China
City	Harbin
Period	1/08/22 → 3/08/22

Keywords

Head detection
Multiple Object Tracking
Person re-Identification

Access to Document

10.1007/978-3-031-13841-6_16

Cite this

Liu, Z., Liu, H., Ren, W., Chang, H., Shi, Y., Lin, R., & Wu, W. (2022). Multiple Object Tracking by Joint Head, Body Detection and Re-Identification. In H. Liu, W. Ren, Z. Yin, L. Liu, L. Jiang, G. Gu, & X. Wu (Eds.), Intelligent Robotics and Applications - 15th International Conference, ICIRA 2022, Proceedings (pp. 171-180). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13458 LNAI). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-13841-6_16

Liu, Zuode ; Liu, Honghai ; Ren, Weihong et al. / Multiple Object Tracking by Joint Head, Body Detection and Re-Identification. Intelligent Robotics and Applications - 15th International Conference, ICIRA 2022, Proceedings. editor / Honghai Liu ; Weihong Ren ; Zhouping Yin ; Lianqing Liu ; Li Jiang ; Guoying Gu ; Xinyu Wu. Springer Science and Business Media Deutschland GmbH, 2022. pp. 171-180 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{af158b0839f2466db901b67354b9b44b,

title = "Multiple Object Tracking by Joint Head, Body Detection and Re-Identification",

abstract = "Multi-object tracking (MOT) is an important problem in computer vision which has a wide range of applications. Formulating MOT as multi-task learning of object detection and re-Identification (re-ID) in a single network is appealing since it achieves real-time but effective inference on detection and tracking. However, in crowd scenes, the existing MOT methods usually fail to locate occluded objects, which also results in bad effects on the re-ID task. To solve people tracking in crowd scenes, we present a model called HBR (Head-Body-ReID Joint Tracking) to jointly formulates head detection, body detection and re-ID tasks into an uniform framework. Human heads are hardly affected by occlusions in crowd scenes, and they can provide informative clues for whole body detection. The experimental results on MOT17 and MOT20 show that our proposed model performs better than the state-of-the-arts.",

keywords = "Head detection, Multiple Object Tracking, Person re-Identification",

author = "Zuode Liu and Honghai Liu and Weihong Ren and Hui Chang and Yuhang Shi and Ruihan Lin and Wenhao Wu",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 15th International Conference on Intelligent Robotics and Applications, ICIRA 2022 ; Conference date: 01-08-2022 Through 03-08-2022",

year = "2022",

doi = "10.1007/978-3-031-13841-6\_16",

language = "英语",

isbn = "9783031138409",

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

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "171--180",

editor = "Honghai Liu and Weihong Ren and Zhouping Yin and Lianqing Liu and Li Jiang and Guoying Gu and Xinyu Wu",

booktitle = "Intelligent Robotics and Applications - 15th International Conference, ICIRA 2022, Proceedings",

address = "德国",

}

Liu, Z, Liu, H , Ren, W, Chang, H, Shi, Y, Lin, R & Wu, W 2022, Multiple Object Tracking by Joint Head, Body Detection and Re-Identification. in H Liu, W Ren, Z Yin, L Liu, L Jiang, G Gu & X Wu (eds), Intelligent Robotics and Applications - 15th International Conference, ICIRA 2022, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13458 LNAI, Springer Science and Business Media Deutschland GmbH, pp. 171-180, 15th International Conference on Intelligent Robotics and Applications, ICIRA 2022, Harbin, China, 1/08/22. https://doi.org/10.1007/978-3-031-13841-6_16

Multiple Object Tracking by Joint Head, Body Detection and Re-Identification. / Liu, Zuode; Liu, Honghai ; Ren, Weihong et al.
Intelligent Robotics and Applications - 15th International Conference, ICIRA 2022, Proceedings. ed. / Honghai Liu; Weihong Ren; Zhouping Yin; Lianqing Liu; Li Jiang; Guoying Gu; Xinyu Wu. Springer Science and Business Media Deutschland GmbH, 2022. p. 171-180 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13458 LNAI).

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

TY - GEN

T1 - Multiple Object Tracking by Joint Head, Body Detection and Re-Identification

AU - Liu, Zuode

AU - Liu, Honghai

AU - Ren, Weihong

AU - Chang, Hui

AU - Shi, Yuhang

AU - Lin, Ruihan

AU - Wu, Wenhao

PY - 2022

Y1 - 2022

N2 - Multi-object tracking (MOT) is an important problem in computer vision which has a wide range of applications. Formulating MOT as multi-task learning of object detection and re-Identification (re-ID) in a single network is appealing since it achieves real-time but effective inference on detection and tracking. However, in crowd scenes, the existing MOT methods usually fail to locate occluded objects, which also results in bad effects on the re-ID task. To solve people tracking in crowd scenes, we present a model called HBR (Head-Body-ReID Joint Tracking) to jointly formulates head detection, body detection and re-ID tasks into an uniform framework. Human heads are hardly affected by occlusions in crowd scenes, and they can provide informative clues for whole body detection. The experimental results on MOT17 and MOT20 show that our proposed model performs better than the state-of-the-arts.

AB - Multi-object tracking (MOT) is an important problem in computer vision which has a wide range of applications. Formulating MOT as multi-task learning of object detection and re-Identification (re-ID) in a single network is appealing since it achieves real-time but effective inference on detection and tracking. However, in crowd scenes, the existing MOT methods usually fail to locate occluded objects, which also results in bad effects on the re-ID task. To solve people tracking in crowd scenes, we present a model called HBR (Head-Body-ReID Joint Tracking) to jointly formulates head detection, body detection and re-ID tasks into an uniform framework. Human heads are hardly affected by occlusions in crowd scenes, and they can provide informative clues for whole body detection. The experimental results on MOT17 and MOT20 show that our proposed model performs better than the state-of-the-arts.

KW - Head detection

KW - Multiple Object Tracking

KW - Person re-Identification

UR - https://www.scopus.com/pages/publications/85137001866

U2 - 10.1007/978-3-031-13841-6_16

DO - 10.1007/978-3-031-13841-6_16

M3 - 会议稿件

AN - SCOPUS:85137001866

SN - 9783031138409

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

SP - 171

EP - 180

BT - Intelligent Robotics and Applications - 15th International Conference, ICIRA 2022, Proceedings

A2 - Liu, Honghai

A2 - Ren, Weihong

A2 - Yin, Zhouping

A2 - Liu, Lianqing

A2 - Jiang, Li

A2 - Gu, Guoying

A2 - Wu, Xinyu

PB - Springer Science and Business Media Deutschland GmbH

T2 - 15th International Conference on Intelligent Robotics and Applications, ICIRA 2022

Y2 - 1 August 2022 through 3 August 2022

ER -

Liu Z, Liu H , Ren W, Chang H, Shi Y, Lin R et al. Multiple Object Tracking by Joint Head, Body Detection and Re-Identification. In Liu H, Ren W, Yin Z, Liu L, Jiang L, Gu G, Wu X, editors, Intelligent Robotics and Applications - 15th International Conference, ICIRA 2022, Proceedings. Springer Science and Business Media Deutschland GmbH. 2022. p. 171-180. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-13841-6_16