基于改进YOLOv11n的无人机图像小目标检测算法OA

In view of the excessive small objects,complex backgrounds,large area of occlusion and limited computing resources in aerial images of unmanned aerial vehicles(UAVs),this paper proposes an algorithm DCC-YOLOv11n based on improved YOLOv11n.Firstly,an improved dilated convolution module C3k2 is introduced into this model to enhance the effect of high-level feature extraction,expand the receptive field,and generate feature maps with richer feature information.Secondly,a context-guided module is introduced into the model to suppress background noise and improve the model 's focus on target regions.Thirdly,inspired by the cross-channel feature modulation(CCFM)mechanism,a channel-adaptive module is integrated into the feature fusion stage to enhance both the model's cross-channel information interaction capability and nonlinear representation capability.Experiments demonstrate that,compared with the baseline YOLOv11n,DCC-YOLOv11n reduces the number of parameters by 8.8%on the VisDrone2019 dataset,with a compact model size of only 5.2 MB.While maintaining a lightweight design,this model sees improvements in precision,recall,mAP@0.5,and mAP@0.5:0.95 of 2.8%,2.6%,2.6%,and 1.8%,respectively,achieves 1.9%and 0.7%gains in mAP@0.5 and mAP@0.5:0.95,respectively,on the HIT-UAV dataset,and a 0.8%improvement in mAP@0.5 on the FOTL_Drone dataset.The above fully validates the effectiveness and generalizability of the proposed model.Overall,DCC-YOLOv11n significantly alleviates the issues of false positives and missed detections commonly observed in traditional UAV-based object detection methods,so it has a certain practicability.