Current vehicle detection methods have leakage problems when performing detection tasks in urban road traffic scenarios with a high percentage of small target vehicles, this paper proposes the YOLOv5s-CB vehicle detection algorithm, which improves the performance of the model for small target vehicle detection by adding the CBAM attention mechanism in the Backbone structure of YOLOv5s and using the BIFPN feature fusion module to replace the Neck layer of the YOLOv5s model. feature fusion module to improve the model's detection performance for small target vehicles. According to the results of the contrast experiment, it can be known that compared with the benchmark algorithm the precision rate by 1.6% the recall rate by 8.1% compared with the YOLOv5s algorithm, mAP@0.5 by 5.2%, which proves that the proposed modification algorithm of our paper makes an efficient improvement in the ability of the vehicle detection model to detect small target vehicles, for vehicle detecting in urban traffic, it is more suitable.
Small Target Vehicle Detection Algorithm Based on YOLOv5s
19.07.2024
1490980 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Vehicle Target Detection Algorithm Based on CBAM-YOLOv5s
IEEE | 2024
|UAV target detection algorithm based on improved YOLOv5s
SPIE | 2023
|