Author(s)

An Chang¹, Dan Li^1*, GanFeng Cao¹, WeiWei Liu², LiKai Wang², Nan Zhou²

Affiliation(s)

¹Xuzhou University of Technology, Xuzhou 221000, Jiangsu, China.

²Traffic police detachment of Xuzhou Public Security Bureau, Xuzhou 221000, Jiangsu, China.

Corresponding Author

Dan Li

ABSTRACT

The purpose of this study is to propose a deep learning-based solution, aiming at the problem of insufficient accuracy and real-time performance in traffic target tracking and trajectory prediction technology. We used YOLOv8 for real-time target detection, combined with the multi-target tracking track algorithm to achieve accurate tracking of traffic targets. At the same time, the trajectory prediction through the long-and short-term memory network (LSTM) can effectively deal with the dynamic changes of traffic flow. The experimental results show that the method tracks better than conventional algorithms in multiple traffic environments, with better robustness and real-time performance. Moreover, this study explores the impact of data enhancement and hyperparameter optimization on model performance, which provides new ideas and methods for the implementation of intelligent transportation system.

KEYWORDS

YOLOv8; Traffic target tracking; Trajectory prediction; Multi-target tracking; Long-and short-term memory network (LSTM)

CITE THIS PAPER

An Chang, Dan Li, GanFeng Cao, WeiWei Liu, LiKai Wang, Nan Zhou. Research on traffic object tracking and trajectory prediction technology based on deep learning. World Journal of Information Technology. 2024, 2(2): 40-47. DOI: 10.61784/wjit3004.

REFERENCES

[1] Kejia Zhu, Yikuan Zhang, Qiang Liu, et al. Research on Application Method of Object Detection in Video. 2022 International Conference on Cloud Computing, Big Data and Internet of Things. 2022, 10, 22.

[2] Li Liu, Wanli Ouyang, Xiaogang Wang, et al. Deep Learning for Generic Object Detection: A Survey. International Journal of Computer Vision, 2020.

[3] Zhengxia Zou, Zhenwei Shi, Yuhong Guo, et al. Object Detection in 20 Years: A Survey. CoRR, 2019.

[4] Kequan Li, Yan Chen, Jiachen Liu, et al. Overview of object detection algorithms based on deep learning. Computer Engineering, 2022, 48(7): 1-12.

[5] Ghania Zidani, Djalal Djarah, Abdslam Benmakhluof, et al. Optimizing Pedestrian Tracking for Robust Perception with YOLOv8 and Deepsort. Applied Computer Science, 2024, 20(1): 72-84.

[6] Shuxin Yang, Yang Xie, Jianqing Wu, et al. CFE-YOLOv8s: Improved YOLOv8s for Steel Surface Defect Detection. ELECTRONICS, 2024, 13(14).

[7] Mukaram Safaldin, Nizar Zaghden, Mahmoud Mejdoub. An Improved YOLOv8 to Detect Moving Objects. IEEE Access, 2024, 12: 59782-59806.

[8] Ji-Yuan Ding, Wang-su Jeon, Sang-Yong Rhee, et al. DM-YOLOv8: Improved Cucumber Disease and Insect Detection Model Based on YOLOV8. 2024.

[9] Shuai Xu, Jiyou Fei, Geng Zhao, et al. CCL-YOLO: Catenary Components Location Based on YOLO and Gather-Distribute Mechanism. IEEE Access, 2024: 1.

[10] Jing Li, Zhengjun Xu, Liang Xu. Vehicle and Pedestrian Detection Method Based on Improved YOLOv4-tiny. Optoelectronics Letters, 2023, 19 (10): 623-628.