Object re-identification method using multi-object tracking Occlusion Area

doi:10.22716/sckt.2025.13.4.003

All Issue

2025 Vol.13, Issue 4 Preview Page Next Page

Research Article

Object re-identification method using multi-object tracking Occlusion Area 사각지역에서의 다중 객체 추적 기술을 활용한 객체 재인식 방법: Sungjun Park¹
박성준¹; ¹ Department of MediaSoftware Engineering, SungKyul University

¹ 성결대학교 미디어소프트웨어학과

31 December 2025. pp. 31-41

PDF

Abstract

본 논문에서는 기존의 다중객체추적 알고리즘을 사용하여 사각 지역에서의 객체 탐지 기술에 대해 새로운 방법을 제안한다. 다중 객체 추적 기술은 실시간 영상을 기반으로 여러 객체들을 연속적으로 추적하는 중요한 알고리즘이다. 그러나, 실제 환경에서는 사각지대의 문제가 발생하여 추적이 단절되고 동일 객체를 지속적으로 탐지가 어렵다. 기존의 연구들은 재식별 기법을 사용하여 해결 방법을 사용하고 있으나, 높은 연산량 등으로 실시간 처리가 어렵다. 본 논문에서는 이러한 문제를 해결하기 위해 비 매칭 보관 기법을 사용한 새로운 방법을 제안한다. IoU 매칭에 실패한 탐지 객체들을 일정시간 보관하여 재 탐지하는 방법이다. 본 연구에서 제안하는 방법을 구현하기 위해서는 YOLO 기반의 객체 검출과 SORT 알고리즘이 기반이 되어야 한다. 본 논문에서는 SORT 알고리즘을 구현하기 위한 칼만 필터 적용 방법, IoU 매칭 알고리즘을 통한 비매칭 보관 방법등에 대해 논의한다. 실험 방법으로는 다중 객체 추적에 대한 정확성과 버퍼의 크기에 따른 성능 실험이다. 실험 결과, 제안된 방법은 기존 다중객체추적 방법들에 비해서 FPS 성능이 우수하였다. 특히, 사각지대에서 객체가 일시적으로 사라졌다가 재등장하는 상황에서 지속적인 추적이 가능한 결과를 얻었다.

This paper proposes a new method for object detection in occlusion using existing multi-object tracking algorithms. Multi-object tracking is a important algorithm for continuously tracking multiple objects based on real-time video. However, in real-world environments, occlusion can occur, disrupting tracking and making it difficult to continuously detect the same object. In older study, have used re-identification techniques to solve this problem, but their high overhead computational real-time processing. In this paper, we suggest a new method using a non-match retention technique. This method retains detected objects that fail IoU matching for a certain period of time and then re-detects them. The proposed method relies on YOLO-based object detection and a SORT algorithm. We discusses Kalman filtering for implementing the SORT algorithm and a non-match retention method using the IoU matching algorithm. Experimental tests include performance tests on the accuracy of multi-object tracking and buffer size. Experimental results show that the proposed method outperforms existing multi-object tracking methods in terms of frames per second (FPS). In particular, we obtained results that enabled continuous tracking in situations where objects temporarily disappeared and then reappeared in occlusion area.

Keywords

Occlusion Area

MOT

SORT

Realtime tracking

Re-Identification

References

P. Voigtlaender, M. Krause, A. Osep, J. Luiten, B. B. G. Sekar, A. Geiger, and B. Leibe, “MOTS: Multi-Object Tracking and Segmentation”, CVPR, pp. 7942-7951, 2019.
10.1109/CVPR.2019.00813
D. Meimetis, I. Daramouskas, I. Perikos, and I. Hatzilygeroudis, “Real-time multiple object tracking using deep learning methods”, Neural Computing and Applications, Vol. 35, No. 1, pp. 89-118, 2023. http://doi.org/10.1007/s00521-021-06391-y
10.1007/s00521-021-06391-y
Y.-B. Park, and Y.-J. Song, “A Method for Detecting Unexpected Situations from Highway Driving Images using Object Detection and Tracking”, JKIIT, Vol. 20, No. 5, pp. 47-54, 2022. http://dx.doi.org/10.14801/jkiit.2022.20.5.47
10.14801/jkiit.2022.20.5.47
J.-H. Song, J.-H. Bai, and S.-H., Park, “Developing a school safety walking monitoring system applied YOLO and DeepSORT model”, JFMSE, Vol. 35, No. 2, pp. 301-314, 2023. https://doi.org/10.13000/JFMSE.2023.4.35.2.301
10.13000/JFMSE.2023.4.35.2.301
C.-H. Seo, N. C. Quy, D. Hu, and S.-R. Chai, “Analysis of multiple object tracking technology trends”, Institute of Control, Robotics and Systems, Vol. 30, No. 2, pp. 13-24, 2024.
J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You only look once : unified, real-time ojbect dectection“, CVPR 2016. doi:10.1109/CVPR.2016.91
10.1109/CVPR.2016.91
M. Kang, and Y.-C. Lim, “Multiple-object tracking method robust to occluded pedestrian tracking”, KSAE, 2021.
https://docs.ultralytics.com, https://github.com/ultralytics/ultralytics
F. Zeong, B. Dong, Y. Zhang, Y. Zhang, T. Wang, X. Zhang, and Y. Wei, “MOTR: End-to-end multiple-object tracking with transformer”, arXiv:2105.03247v4, 2022. https://github.com/megvii-research/MOTR
10.1007/978-3-031-19812-0_38
J. Oh, and J. Lee, “Trends on visual object tracking using siamese network”, ETRI, Vol. 37, No. 1, pp. 73-83, 2022. https://doi.org/10.22648/ETRI.2022.J.370108
10.22648/ETRI.2022.J.370108
M. Kang, and Y.-C. Lim, “Multiple-object tracking method robust to occluded pedestrian tracking”, SKSAE_J008, pp. 919-922.
A. Bewley, Z. Ge, L. Ott, F. Ramos, and B. Upcroft, “Simple online and realtime tracking”, ICIP, pp. 3464-3468, 2016.
10.1109/ICIP.2016.7533003
K. Yi, K. Luo, X. Luo, J. Huang, H. Wu, R. Hu, and W. Hao, “UCMCTrack: multi-object tracking with uniform camera motion compensation”, Computer Vision and Pattern Recognition, 2024. https://doi.org/10.1609/aaai.v38i7.28493
10.1609/aaai.v38i7.28493
M. Yang, G. Han, B. Yan, W. Zhang, J. Qi, H. Lu, and D. Wang, “Hybrid-SORT: cues matter for online multi-object tracking”, arXiv:2308.00783v2, 2024.
10.1609/aaai.v38i7.28471
T. Meinhardt, A. Kirillov, L. Leal-Taixe, and C. Feichtenhofer, “TrackFormer: multi-object tracking with transformers”, CVPR. 2022. https://doi.org/10.1109/CVPR52688.2022.00864
10.1109/CVPR52688.2022.00864

Information

Publisher :The Society of Convergence Knowledge
Publisher(Ko) :융복합지식학회
Journal Title :The Society of Convergence Knowledge Transactions
Journal Title(Ko) :융복합지식학회논문지
Volume : 13
No :4
Pages :31-41
DOI :https://doi.org/10.22716/sckt.2025.13.4.003

[1] P. Voigtlaender, M. Krause, A. Osep, J. Luiten, B. B. G. Sekar, A. Geiger, and B. Leibe, “MOTS: Multi-Object Tracking and Segmentation”, CVPR, pp. 7942-7951, 2019.
10.1109/CVPR.2019.00813

[2] D. Meimetis, I. Daramouskas, I. Perikos, and I. Hatzilygeroudis, “Real-time multiple object tracking using deep learning methods”, Neural Computing and Applications, Vol. 35, No. 1, pp. 89-118, 2023. http://doi.org/10.1007/s00521-021-06391-y
10.1007/s00521-021-06391-y

[3] Y.-B. Park, and Y.-J. Song, “A Method for Detecting Unexpected Situations from Highway Driving Images using Object Detection and Tracking”, JKIIT, Vol. 20, No. 5, pp. 47-54, 2022. http://dx.doi.org/10.14801/jkiit.2022.20.5.47
10.14801/jkiit.2022.20.5.47

[4] J.-H. Song, J.-H. Bai, and S.-H., Park, “Developing a school safety walking monitoring system applied YOLO and DeepSORT model”, JFMSE, Vol. 35, No. 2, pp. 301-314, 2023. https://doi.org/10.13000/JFMSE.2023.4.35.2.301
10.13000/JFMSE.2023.4.35.2.301

[5] C.-H. Seo, N. C. Quy, D. Hu, and S.-R. Chai, “Analysis of multiple object tracking technology trends”, Institute of Control, Robotics and Systems, Vol. 30, No. 2, pp. 13-24, 2024.

[6] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You only look once : unified, real-time ojbect dectection“, CVPR 2016. doi:10.1109/CVPR.2016.91
10.1109/CVPR.2016.91

[7] M. Kang, and Y.-C. Lim, “Multiple-object tracking method robust to occluded pedestrian tracking”, KSAE, 2021.

[8] https://docs.ultralytics.com, https://github.com/ultralytics/ultralytics

[9] F. Zeong, B. Dong, Y. Zhang, Y. Zhang, T. Wang, X. Zhang, and Y. Wei, “MOTR: End-to-end multiple-object tracking with transformer”, arXiv:2105.03247v4, 2022. https://github.com/megvii-research/MOTR
10.1007/978-3-031-19812-0_38

[10] J. Oh, and J. Lee, “Trends on visual object tracking using siamese network”, ETRI, Vol. 37, No. 1, pp. 73-83, 2022. https://doi.org/10.22648/ETRI.2022.J.370108
10.22648/ETRI.2022.J.370108

[11] M. Kang, and Y.-C. Lim, “Multiple-object tracking method robust to occluded pedestrian tracking”, SKSAE_J008, pp. 919-922.

[12] A. Bewley, Z. Ge, L. Ott, F. Ramos, and B. Upcroft, “Simple online and realtime tracking”, ICIP, pp. 3464-3468, 2016.
10.1109/ICIP.2016.7533003

[13] K. Yi, K. Luo, X. Luo, J. Huang, H. Wu, R. Hu, and W. Hao, “UCMCTrack: multi-object tracking with uniform camera motion compensation”, Computer Vision and Pattern Recognition, 2024. https://doi.org/10.1609/aaai.v38i7.28493
10.1609/aaai.v38i7.28493

[14] M. Yang, G. Han, B. Yan, W. Zhang, J. Qi, H. Lu, and D. Wang, “Hybrid-SORT: cues matter for online multi-object tracking”, arXiv:2308.00783v2, 2024.
10.1609/aaai.v38i7.28471

[15] T. Meinhardt, A. Kirillov, L. Leal-Taixe, and C. Feichtenhofer, “TrackFormer: multi-object tracking with transformers”, CVPR. 2022. https://doi.org/10.1109/CVPR52688.2022.00864
10.1109/CVPR52688.2022.00864

The Society of Convergence Knowledge Transactions ISSN:2287-8920(Print) 융복합지식학회논문지

All Issue