基于改进YOLOv11的无人机雾天车辆目标检测算法研究OA

Since unmanned aerial vehicles(UAVs)often suffer low multi-scale target detection accuracy,insufficient feature contrast and limited computing resources caused by image degradation in foggy scenes,this paper proposes a lightweight foggy vehicle detection algorithm YOLO-RDPF based on im-proved YOLOv11.First,the C3K2_RCB module and the DWConv module are introduced into the back-bone network to enhance its feature extraction ability.Second,the SimAM module is incorporated to high-light key features and suppress redundant information.Finally,a feature focusing dispersion feature pyra-mid network(FDFPN)is constructed to further improve the multi-scale feature fusion and representation.The experimental results on the HazyDet dataset show that the YOLO-RDPF algorithm achieves the mAP50 and mAP50-95 of 3.1%and 2.5%,respectively,which are both higher than those of the YOLOv11n algorithm.The mAP50 reaches 68.3%,while the number of model parameters is only 1.8×106.These re-sults demonstrate that YOLO-RDPF can significantly improve the vehicle detection performance in foggy scenes while maintaining a high degree of lightweight.