Embedding with Bounding Box Contracting for Multi-object Tracking

Like Zhang; Wenjing Kang; Guangdong Zhang

doi:10.1007/978-3-031-36011-4_6

Embedding with Bounding Box Contracting for Multi-object Tracking

Like Zhang
, Wenjing Kang^*
, Guangdong Zhang

^*Corresponding author for this work

School of Information Science and Engineering, Harbin Institute of Technology Weihai

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

Abstract

The development of 5G/6G network can achieve high data transmission speed, which promotes the wide application of remote video monitoring. Multi-object tracking (MOT) aims at detecting and tracking all the objects of interesting categories in videos. Appearance and motion information of each object are significant clues utilized for finding associations between detections and tracks. Many approaches model each object appearance through bounding box region, which is vulnerable to background noise and motion deformation. In this paper, we alleviate this problem, via embedding with object bounding box contracting. We also integrate an online tracking by detection model, comprehensive use of appearance and motion information for data association. Object bounding box contracting is introduced to relieve the impact of interference and obtain high-quality re-ID embeddings. Experimental results based on the MOT17 benchmark show that the integrated tracker with bounding box contracting for embedding achieves 80.6 MOTA, 79.4 IDF1 and 64.4 HOTA.

Original language	English
Title of host publication	6GN for Future Wireless Networks - 5th EAI International Conference, 6GN 2022, Proceedings
Editors	Ao Li, Liang Xi, Yao Shi
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	65-76
Number of pages	12
ISBN (Print)	9783031360107
DOIs	https://doi.org/10.1007/978-3-031-36011-4_6
State	Published - 2023
Externally published	Yes
Event	5th EAI International Conference on 6G for Future Wireless Networks, 6GN 2022 - Harbin, China Duration: 17 Dec 2022 → 18 Dec 2022

Publication series

Name	Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST
Volume	504 LNICST
ISSN (Print)	1867-8211
ISSN (Electronic)	1867-822X

Conference

Conference	5th EAI International Conference on 6G for Future Wireless Networks, 6GN 2022
Country/Territory	China
City	Harbin
Period	17/12/22 → 18/12/22

Keywords

Embedding Methods
Multi-Object Tracking
Object Detecting

Access to Document

10.1007/978-3-031-36011-4_6

Cite this

Zhang, L., Kang, W., & Zhang, G. (2023). Embedding with Bounding Box Contracting for Multi-object Tracking. In A. Li, L. Xi, & Y. Shi (Eds.), 6GN for Future Wireless Networks - 5th EAI International Conference, 6GN 2022, Proceedings (pp. 65-76). (Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST; Vol. 504 LNICST). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-36011-4_6

Zhang, Like ; Kang, Wenjing ; Zhang, Guangdong. / Embedding with Bounding Box Contracting for Multi-object Tracking. 6GN for Future Wireless Networks - 5th EAI International Conference, 6GN 2022, Proceedings. editor / Ao Li ; Liang Xi ; Yao Shi. Springer Science and Business Media Deutschland GmbH, 2023. pp. 65-76 (Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST).

@inproceedings{234d9a7bc9e44f5d89353c6ea18149e7,

title = "Embedding with Bounding Box Contracting for Multi-object Tracking",

abstract = "The development of 5G/6G network can achieve high data transmission speed, which promotes the wide application of remote video monitoring. Multi-object tracking (MOT) aims at detecting and tracking all the objects of interesting categories in videos. Appearance and motion information of each object are significant clues utilized for finding associations between detections and tracks. Many approaches model each object appearance through bounding box region, which is vulnerable to background noise and motion deformation. In this paper, we alleviate this problem, via embedding with object bounding box contracting. We also integrate an online tracking by detection model, comprehensive use of appearance and motion information for data association. Object bounding box contracting is introduced to relieve the impact of interference and obtain high-quality re-ID embeddings. Experimental results based on the MOT17 benchmark show that the integrated tracker with bounding box contracting for embedding achieves 80.6 MOTA, 79.4 IDF1 and 64.4 HOTA.",

keywords = "Embedding Methods, Multi-Object Tracking, Object Detecting",

author = "Like Zhang and Wenjing Kang and Guangdong Zhang",

note = "Publisher Copyright: {\textcopyright} 2023, ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering.; 5th EAI International Conference on 6G for Future Wireless Networks, 6GN 2022 ; Conference date: 17-12-2022 Through 18-12-2022",

year = "2023",

doi = "10.1007/978-3-031-36011-4\_6",

language = "英语",

isbn = "9783031360107",

series = "Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST",

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

pages = "65--76",

editor = "Ao Li and Liang Xi and Yao Shi",

booktitle = "6GN for Future Wireless Networks - 5th EAI International Conference, 6GN 2022, Proceedings",

address = "德国",

}

Zhang, L, Kang, W & Zhang, G 2023, Embedding with Bounding Box Contracting for Multi-object Tracking. in A Li, L Xi & Y Shi (eds), 6GN for Future Wireless Networks - 5th EAI International Conference, 6GN 2022, Proceedings. Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST, vol. 504 LNICST, Springer Science and Business Media Deutschland GmbH, pp. 65-76, 5th EAI International Conference on 6G for Future Wireless Networks, 6GN 2022, Harbin, China, 17/12/22. https://doi.org/10.1007/978-3-031-36011-4_6

Embedding with Bounding Box Contracting for Multi-object Tracking. / Zhang, Like; Kang, Wenjing; Zhang, Guangdong.
6GN for Future Wireless Networks - 5th EAI International Conference, 6GN 2022, Proceedings. ed. / Ao Li; Liang Xi; Yao Shi. Springer Science and Business Media Deutschland GmbH, 2023. p. 65-76 (Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST; Vol. 504 LNICST).

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

TY - GEN

T1 - Embedding with Bounding Box Contracting for Multi-object Tracking

AU - Zhang, Like

AU - Kang, Wenjing

AU - Zhang, Guangdong

PY - 2023

Y1 - 2023

N2 - The development of 5G/6G network can achieve high data transmission speed, which promotes the wide application of remote video monitoring. Multi-object tracking (MOT) aims at detecting and tracking all the objects of interesting categories in videos. Appearance and motion information of each object are significant clues utilized for finding associations between detections and tracks. Many approaches model each object appearance through bounding box region, which is vulnerable to background noise and motion deformation. In this paper, we alleviate this problem, via embedding with object bounding box contracting. We also integrate an online tracking by detection model, comprehensive use of appearance and motion information for data association. Object bounding box contracting is introduced to relieve the impact of interference and obtain high-quality re-ID embeddings. Experimental results based on the MOT17 benchmark show that the integrated tracker with bounding box contracting for embedding achieves 80.6 MOTA, 79.4 IDF1 and 64.4 HOTA.

AB - The development of 5G/6G network can achieve high data transmission speed, which promotes the wide application of remote video monitoring. Multi-object tracking (MOT) aims at detecting and tracking all the objects of interesting categories in videos. Appearance and motion information of each object are significant clues utilized for finding associations between detections and tracks. Many approaches model each object appearance through bounding box region, which is vulnerable to background noise and motion deformation. In this paper, we alleviate this problem, via embedding with object bounding box contracting. We also integrate an online tracking by detection model, comprehensive use of appearance and motion information for data association. Object bounding box contracting is introduced to relieve the impact of interference and obtain high-quality re-ID embeddings. Experimental results based on the MOT17 benchmark show that the integrated tracker with bounding box contracting for embedding achieves 80.6 MOTA, 79.4 IDF1 and 64.4 HOTA.

KW - Embedding Methods

KW - Multi-Object Tracking

KW - Object Detecting

UR - https://www.scopus.com/pages/publications/85169072003

U2 - 10.1007/978-3-031-36011-4_6

DO - 10.1007/978-3-031-36011-4_6

M3 - 会议稿件

AN - SCOPUS:85169072003

SN - 9783031360107

T3 - Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST

SP - 65

EP - 76

BT - 6GN for Future Wireless Networks - 5th EAI International Conference, 6GN 2022, Proceedings

A2 - Li, Ao

A2 - Xi, Liang

A2 - Shi, Yao

PB - Springer Science and Business Media Deutschland GmbH

T2 - 5th EAI International Conference on 6G for Future Wireless Networks, 6GN 2022

Y2 - 17 December 2022 through 18 December 2022

ER -

Zhang L, Kang W, Zhang G. Embedding with Bounding Box Contracting for Multi-object Tracking. In Li A, Xi L, Shi Y, editors, 6GN for Future Wireless Networks - 5th EAI International Conference, 6GN 2022, Proceedings. Springer Science and Business Media Deutschland GmbH. 2023. p. 65-76. (Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST). doi: 10.1007/978-3-031-36011-4_6

Embedding with Bounding Box Contracting for Multi-object Tracking

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this