Research on an Identification System for Illegal Occupation of Bus-only Lanes Based on YOLOv8-Seg and ByteTrack

DOI: https://doi.org/10.62517/jike.202604407

Author(s)

Shengyuan Lei

Affiliation(s)

Chang'an University, Xi'an, Shaanxi, China

Abstract

Traffic violations involving social vehicles illegally occupying bus-only lanes have occurred frequently. Traditional regulatory approaches based on manual monitoring and patrol-based image capture often struggle to ensure detection efficiency and supervision accuracy during peak traffic periods or under adverse weather conditions. Based on computer vision technology and integrated with advanced visual algorithms, this study develops an identification and detection system for illegal occupation of bus-only lanes by motor vehicles.The dataset used in this study was obtained from videos captured by fixed roadside devices. During the data preprocessing stage, bus lanes and vehicles were annotated, and data augmentation strategies such as instance paste were employed to improve dataset quality. To achieve end-to-end synchronous detection of vehicle positions and lane areas, the YOLOv8-Seg model was adopted for training and detection. To establish a scientifically sound and compliant violation judgment mechanism, this study introduced the ByteTrack high-precision vehicle target tracking algorithm and integrated bottom projection, the ray casting method, and a time-threshold-based state machine to determine whether a vehicle continuously occupied a bus-only lane, thereby reducing false alarms caused by short-term lane crossing and legal lane changing. Experimental results show that the high-precision visual algorithm constructed in this study ensures the accuracy of identifying illegal occupation of bus-only lanes and effectively reduces the probability of misjudgment. While lowering hardware deployment costs, the proposed approach can assist urban traffic law enforcement and provide a reference for intelligent traffic management and control in modern cities.

Keywords

Intelligent Transportation System; YOLOv8; ByteTrack; Object Detection; Instance Segmentation

References

[1] Chen Meng. Research on a violation detection algorithm for lane occupation based on in-vehicle video footage [D]. Huazhong University of Science and Technology, 2015. [2] Dawood B, Mujtaba Z, Akbari P, et al. Automated enforcement of bus-only lanes and crossings with policy and implementation insights from a calgary case study[J]. Case Studies on Transport Policy, 2025: 101630. [3] Sun Xiyun. Recognition of motor vehicles illegally using bus lanes based on deep learning [D]. China People's Public Security University, 2025 [4] Li Zhiyuan, Wang Guanghui. Research on Lane Line Recognition Based on Traditional Edge Operators [J]. Modern Electronic Technology, 2024,47(07):61–65. [5] Ge Z, Ma C, Fu Z, et al. End-to-end lane detection with convolution and transformer[J]. Multimedia Tools and Applications, 2023, 82(19): 29607-29627. [6] Ren S, He K, Girshick R, et al. Faster r-cnn: Towards real-time object detection with region proposal networks[J]. Advances in neural information processing systems, 2015, 28. [7] JOCHER G, CHAURASIA A, QIU J. Ultralytics YOLOv8: version 8.0.0[CP/OL]. 2023[2026-07-24]. https://github.com/ultralytics/ultralytics. [8] Zhang Y, Sun P, Jiang Y, et al. Bytetrack: Multi-object tracking by associating every detection box[C]//European conference on computer vision. Cham: Springer Nature Switzerland, 2022: 1-21. [9] GB 5768.1-2025: Road Traffic Signs and Markings – Part 1: General Principles [S]. Beijing: China Standards Press, 2025. [10] Qin X, Zhang Z, Huang C, et al. U2-Net: Going deeper with nested U-structure for salient object detection[J]. Pattern recognition, 2020, 106: 107404. [11] Pan S J, Yang Q. A survey on transfer learning[J]. IEEE Transactions on knowledge and data engineering, 2009, 22(10): 1345-1359. [12] Lin T Y, Maire M, Belongie S, et al. Microsoft coco: Common objects in context[C]//European conference on computer vision. Cham: Springer International Publishing, 2014: 740-755. [13] Bolya D, Zhou C, Xiao F, et al. Yolact: Real-time instance segmentation[C]//Proceedings of the IEEE/CVF international conference on computer vision. 2019: 9157-9166.