With the rapid increase in the number of automobiles, urban traffic management is increasingly facing serious challenges, so efficient vehicle detection technology is particularly important. In this paper, we investigate vehicle detection and early warning techniques based on the improved YOLOv8 network. We propose an improved method that combines a deep training strategy, a lightweight model design, and introduces an improved MPDIoU loss function to optimize the model's performance in vehicle detection. The experimental results show that the improved model shows significant improvement in precision, recall and F1 value compared to the pre-improvement, especially in the detection ability in complex environments. Compared with both YOLOv8 and the latest YOLOv9 model, our model achieves better results in terms of performance and shows better application prospects.
Vehicle detection and early warning research based on improved YOLOv8 networks
International Conference on Computer Graphics, Artificial Intelligence, and Data Processing (ICCAID 2024) ; 2024 ; Nanchang, China
Proc. SPIE ; 13560 ; 135602W
2025-04-10
Conference paper
Electronic Resource
English