Research and Analysis of Multi object Tracking Combined with Transformer Method

DOI

10.2991/978-94-6239-648-7_90How to use a DOI?

Keywords

Multi-object tracking; Transformer; Feature fusion; Data association; Occlusion handling

Abstract

Multi-object tracking (MOT) is widely used in intelligent transportation, public safety, and autonomous driving. Traditional MOT algorithms, relying on local feature modeling and heuristic association rules, have reached their performance limits. This leads to significant performance degradation in complex scenarios such as object occlusion, identity switching (IDSW), and scale variations. In recent years, the Transformer architecture, with its global self-attention mechanism and long-term dependency modeling capabilities, has established a new paradigm for MOT, achieving end-to-end collaborative optimization of object detection and identity mapping. This paper provides a comprehensive overview of Transformer-based MOT research. First, it describes the task definition and key evaluation metrics, and compares traditional MOT methods, CNN-based methods, and Transformer-based solutions. Then, it analyzes the core innovations in three key areas: feature extraction, decoder design, and data association. The paper focuses on techniques such as bi-branch feature fusion, spatial location constraints, linear attention optimization, and multi-scale object modeling. Furthermore, this paper utilizes benchmark datasets such as MOT17, MOT20, KITTI, and VisDrone-MOT2019 to compare representative Transformer-based algorithms in terms of tracking accuracy (MOTA, IDF1, HOTA) and real-time performance (frame rate). Finally, this paper summarizes the unresolved challenges and future research directions in current research. The aim of this work is to lay a systematic foundation for the theoretical research and practical applications of Transformer-based MOT.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Junhao Yu
PY  - 2026
DA  - 2026/04/24
TI  - Research and Analysis of Multi object Tracking Combined with Transformer Method
BT  - Proceedings of the International Workshop on Advances in Deep Learning for Image Analysis and Computer Vision (IWADIC 2025)
PB  - Atlantis Press
SP  - 829
EP  - 842
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6239-648-7_90
DO  - 10.2991/978-94-6239-648-7_90
ID  - Yu2026
ER  -