基于YOLOv8n的无人机视角红外小目标检测算法OA

In infrared ground target detection from an unmanned aerial vehicle(UAV),small target detection is plagued by the problems of false detection and missed detection.To tackle this issue,we propose an improved YOLOv8n detection algorithm,which optimizes YOLOv8n in multiple aspects.Firstly,a 160×160 small target detection layer is added,and the feature map output by the first C2f module of the original network is introduced into the detection head to strengthen the detection of small-sized targets.Meanwhile,a dual-convolution CSP_BiFormer bottleneck module is incorporated to enhance the feature extraction capability and effectively handle the long-range dependencies between features.Secondly,the loss function is improved by combining EIoU and CIoU to optimize the adjustment of the length and width of the prediction box.Finally,a multi-scale feature adaptive attention module is introduced to fuse information of different scales,thereby improving the detection performance for both small and large targets.Experimental results on the HIT-UAV dataset show that compared with YOLOv8n,the improved algorithm achieves increases of 10.4 percentage points in precision,3.7 percentage points in recall,6.7 percentage points in mean average precision(mAP@0.5),5.4 percentage points in mAP@0.5:0.95,and 15.4 frames per second in FPS.Moreover,it exhibits better detection performance in complex scenarios compared with current main-stream algorithms,which verifies the effectiveness and good generalization ability of the proposed algorithm.