Research on Multi-source Video Streaming Vehicle Object Detection and Parallel Statistics Based on YOLO-OBB

Qian Jiang

doi:10.63313/AJET.9023

Authors

Qian Jiang School of Automotive and Traffic Engineering, Liaoning University of Technology, Jinzhou 121000, China Author

DOI:

https://doi.org/10.63313/AJET.9023

Keywords:

YOLO-OBB, statistical analysis of multi-intersection vehicles, dual-threshold steering recognition algorithm, multi-threaded processing, PySide6

Abstract

Based on the YOLO-OBB rotating object detection model, a statistical analysis system for multi-intersection vehicles is proposed. The system realizes the accurate analysis of vehicles running in complex intersection scenarios by analyzing the characteristics of the vehicle's driving trajectory and the characteristic change law of the steering angle in the intersection area, and integrating the rotating bounding box detection technology and the innovative dual-threshold steering recognition algorithm. The system can not only accurately identify the entrance direction of the vehicle, but also judge the steering behavior and complete the model classification in real time. To improve efficiency, the system adopts a multi-threaded parallel architecture that can analyze 9 high-definition video streams simultaneously, and establishes an intuitive visual interface through the PySide6 framework. The experimental results show that on the 1920×1080 resolution video stream, the vehicle recognition accuracy of the four intersections is 92.2%, and the accuracy of vehicle and steering judgment is also more than 80%, which can meet the demand for real-time and accurate vehicle flow data in actual scenarios, and provide data support for decisions such as signal light optimization and road planning.

References

[1] Liu Dongwen, Yang Keqiang, Lu Jiaming. Chinese Cities with "Capital Letters"[N]. Manager Daily, 2012-08-31.

[2] Ministry of Public Security of the People's Republic of China. In 2023, the number of mo-tor vehicles in China will reach 435 million, and the total number of drivers will reach 523 million [EB/OL]. (2024-01-11) [2025-10-20].https://www.mps.gov.cn/n6557563/c9386285/content.html

[3] Parsa B A ,Taghipour H ,Derrible S , et al.Real-time accident detection: Coping with imbal-anced data[J].Accident Analysis and Preven-tion,2019,129202-210.DOI:10.1016/j.aap.2019.05.014.

[4] LYU N C，PENG L F，WU C Z，et al. Real-time crash-risk prediction model that distin-guishes collision types [J]. China Journal of Highway and Transport，2022，35（1）：93-108.（in Chinese）

[5] GUAN L M，ZHANG Q，CHU Q L，et al. Roadside dual lidar data fusion based on im-proved ICP algorithm[J]. Laser Journal，2021，42（9）：38-44（. in Chinese）

[6] XUE Q W，JIANG Y M，LU J. Risky driving behavior recog90 nition based on trajectory data[J]. China Journal of Highway and Transport，2020，33（6）：84-94（. in Chinese）

[7] JOCHER G, STOKEN A, CHAURASIA A, et al. Ultralytics/YOLOv5: v6.0 - YOLOv5n 'Nano' models, Roboflow integration, TensorFlow export, OpenCV DNN support[Z/OL]. (2021-10-29) [2025-10-20]. https://doi.org/10.5281/zenodo.5563715.

[8] REN S Q,HE K M,GIRSHICK R,et al.Faster R-CNN:Towards Real-Time Object Detection with Region Proposal Networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelli-gen ce,2017,39(6):1137-1149.

[9] Girshick R. Fast R-CNN[C]//Proceedings of the IEEE International Conference on Comput-er Vision（ICCV）,Santiago,Chile,2015:1440-1448.

[10] Liu Chao, Luo Ruyi, Liu Chunqing, et al. Vehicle Continuous Trajectory Construction Meth-od Based on Roadside Multi-camera Video Target Correlation and Trajectory Stitch-ing[J].Traffic Information and Safe-ty,2023,41(03):80-91.DOI:CNKI:SUN:JTJS.0.2023-03-009.

[11] Jocher, G., Chaurasia, A. and Qiu, J. (2023) YOLO by Ultralytics. https://github.com/ultralytics/ultralytics Guo S, Li W, Zhang Q, et al. Vehicle pose estima-tion for turn detection in complex intersections[J]. IEEE Robotics and Automation Letters, 2024, 9(2): 1567-1574.