Summary of Multi-object Tracking Methods Combined with Deep Learning
DOI:
https://doi.org/10.54097/2efrm195Keywords:
MOT; Computer vision; Deep learning.Abstract
The essence of the multi-target tracking (MOT) task is to identify and position multiple targets in the video sequence and ensure the continuity of the target identity in the time dimension, so as to produce the comprehensive path of each target in the whole video. Due to advancements in deep learning, multi-target tracking technology has undergone substantial enhancement, and is applicable to a variety of realistic application scenarios, such as autonomous driving, video surveillance, and action analysis, to meet diverse needs. On the basis of extensive literature research, according to the multi-target tracking algorithm framework, the current mainstream multi-target tracking algorithm is divided into multi-target tracking, joint detection, multi-target tracking and end-to-end multi-target tracking based on attention mechanism and introduce all kinds of representative algorithms in the framework of feature embedding. This paper examines the technical characteristics, advantages and disadvantages, and applicable environment of each type of algorithm in detail and shows the experimental results of each algorithm on the standard dataset MOT 17, so as to aid researchers in selecting suitable models for research.
Downloads
References
[1] H. Liu, Multi-object Tracking Algorithm Based on Deep Learning, Jiangnan University, DOI:10.27169/d.cnki.gwqgu, 002649 (2023).
[2] X. Bai, Research on multi-object tracking method, Taiyuan University of Science and Technology, DOI:10.27721/d.cnki.gyzjc, 000549 (2023).
[3] F. Du, Research on the visual target tracking method based on the attention mechanism, And Harbin Institute of Technology, DOI:10.27061/d.cnki.ghgdu, 000239 (2021) .
[4] A. Bewley, Z. Ge, L. Ott, et al, Simple online and realtime tracking//2016 IEEE international conference on image processing (ICIP), IEEE, 3464-3468 (2016).
[5] N. Wojke, A. Bewley, D. Paulus, Simple online and realtime tracking with a deep association metric, 2017 IEEE international conference on image processing (ICIP), 3645-3649 (2017).
[6] Y. Zhang, P. Sun, Y. Jiang, et al, Bytetrack: Multi-object tracking by associating every detection box//European conference on computer vision, Cham: Springer Nature Switzerland, 1-21 (2022).
[7] Z. Wang, L. Zheng, Y. Liu, et al, towards real-time multi-object tracking//European conference on computer vision, Cham: Springer International Publishing, 107-122 (2020).
[8] Y. Zhang, C. Wang, X. Wang, et al, Fairmot: On the fairness of detection and re-identification in multiple object tracking. International journal of computer vision. 129, 3069-3087(2021).
[9] X. Zhou, V. Koltun, P. Krähenbühl, Tracking objects as points//European conference on computer vision, Cham: Springer International Publishing, 474-490 (2020).
[10] C. Liang, Z. Zhang, X. Zhou, et al, Rethinking the competition between detection and reid in multiobject tracking. IEEE Transactions on Image Processing. 31, 3182-3196(2022).
[11] T. Meinhardt, A. Kirillov, L. Leal-Taixe, et al, Trackformer: Multi-object tracking with transformers//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 8844-8854 (2022).
[12] P. Sun, J. Cao, Y. Jiang, et al, Transtrack: Multiple object tracking with transformer, arXiv preprint arXiv: 2012, 15460 (2020).
[13] P. Chu, J. Wang, Q. You, et al, Transmot: Spatial-temporal graph transformer for multiple object tracking//Proceedings of the IEEE/CVF Winter Conference on applications of computer vision, 4870-4880 (2023).
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Highlights in Science, Engineering and Technology
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
How to Cite
