Research on Multi-target Vehicle Tracking Algorithm Based on Improved YOLOv8+DeepSORT
(School of Computer Science, Xi'an Shiyou University, Xi'an 710065, China)
Abstract
Aiming at the issues that vehicle tracking algorithms are highly complex and computationally intensive in practical traffic scenarios, making it difficult to apply them to devices with limited resources, this paper proposes a multi-target vehicle tracking algorithm based on improved YOLOv8+DeepSORT. A lightweight GhostNet Convolutional Neural Network is introduced into the backbone network, and the Conv is replaced with GhostConv. This replacement not only ensures the lightweight nature of the model but also improves its performance. Subsequently, by introducing the CBAM and integrating it with the Ghost convolution technology, a new GC-C2f feature fusion module is constructed to further enhance the feature extraction ability. Finally, a new loss function, WIoU, is adopted to improve the model's regression accuracy and convergence speed. The detection results of the improved YOLOv8 model are used as the input of the DeepSORT algorithm to achieve multi-target vehicle tracking in complex situations. Experimental results demonstrate that on the KITTI traffic dataset, without sacrificing detection accuracy, compared with the original YOLOv8+DeepSORT, the parameter count of GCW-YOLO+DeepSORT is reduced by 35.94% and the computational load is decreased by 20.25%. This makes it more suitable for deployment on devices with limited resources and endows it with practical value.
Keywords
vehicle target tracking; YOLOv8; DeepSORT; Ghost convolution; Lightweight type; CBAM; Loss Function
References
- [1] GIRSHICK R,DONAHUE J,DARRELL T,et al. Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation [C]//2014 IEEE Conference on Computer Vision and Pattern Recognition.Columbus:IEEE,2014:580-587.
- [2] Cai Xiumei, Wang Yan, Bian Jingwei, etc A review of multi-target tracking data association algorithms [J]. Journal of Xi'an University of Posts and Telecommunications, 2021, 26 (2): 77-86
- [3] REN S,HE K,GIRSHICK R,et al. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks [J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149.
- [4] BOCHKOVSKIY A,WANG C-Y,LIAO H-Y M. YOLOv4: Optimal Speed and Accuracy of Object Detection [J/OL].arXiv:2004.10934 [cs.CV].[2024-12-03].https://arxiv.org/abs/2004.10934.
- [5] Pu Lingling, Yangliu Improving YOLOv5's real-time detection and tracking algorithm for multi vehicle targets [J]. Science, Technology and Engineering, 2023, 23 (28): 12159-12167
- [6] Li Mengkun, Yuan Chen, Wang Qi, etc Research on Online Listening Behavior Recognition Model Based on Improved YOLOv8 Algorithm [J]. Computer Engineering, 2025, 51 (1): 287-294
- [7] Zhang Lifeng, Tian Ying Improved multi-scale lightweight vehicle object detection algorithm for YOLOv8 [J]. Computer Engineering and Applications, 2024, 60 (3): 129-137
- [8] HAN K,WANG Y,TIAN Q,et al. GhostNet: More Features from Cheap Operations [C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Seattle:IEEE,2020:1580-1589.
- [9] WOO S,PARK J,LEE J Y,et al. CBAM: Convolutional Block Attention Module [C]//Proceedings of the European Conference on Computer Vision (ECCV).Munich:Springer,2018:3-19.
- [10] TONG Z,CHEN Y,XU Z,et al. Wise-IoU: Bounding Box Regression Loss with Dynamic Focusing Mechanism [J/OL].arXiv:2301.10051 [cs.CV].[2024-12-05].https://arxiv.org/abs/2301.10051.
- [11] REDMON J,FARHADI A. YOLOv3: An Incremental Improvement [J/OL].arXiv:1804.02767 [cs.CV].[2024-12-03].https://arxiv.org/pdf/1804.02767.
- [12] ZHANG Y,WANG C,WANG X,et al. FairMOT: On the Fairness of Detection and Re-Identification in Multiple Object Tracking [J].International Journal of Computer Vision,2021,129:3069-3087.
- [13] GEIGER A,LENZ P,STILLER C,et al. Vision Meets Robotics: The KITTI Dataset [J].The International Journal of Robotics Research,2013,32(11):1231-1237.
About the authors
Xu Yingzhuo (1964-), female, Han ethnicity, from Xi'an, Shaanxi Province, second level professor, graduate supervisor, master's degree, research direction: deep learning and artificial intelligence, program design; Corresponding author: Zhang Yuanbo (1998-), male, Han ethnicity, from Luoyang, Henan Province, currently pursuing a master's degree in computer vision and deep learning; Liu Xiaotian (1999-), male, Han ethnicity, from Jinan, Shandong, is currently pursuing a master's degree in computer vision and deep learning.