基于Swin Transformer和YOLOv8的玉米叶病害识别算法研究OA

In order to improve the accuracy of identifying corn leaf disease pests,this paper proposed an improved algorithm that com-bines Swin Transformer and YOLOv8 network.Based on the YOLOv8 network,modules such as Focus and Depthwise Convolution were introduced to reduce computation and parameters,increase the receptive field and feature channels,and improve feature fusion and transmission capabilities.Additionally,the Wise Intersection over Union(WIoU)loss function was adopted to optimize the network.The experimental results showed that the Swin Transformer-YOLO model achieved excellent performance on the self-built corn leaf dis-ease dataset,with an accuracy of 91.5%and a mean average precision(mAP@0.5)of 89.4%,significantly outperforming other detec-tors.Compared to mainstream algorithms(such as YOLOv8,YOLOv7,YOLOv5,and YOLOx),the Swin Transformer-YOLO model ex-celled in all metrics,particularly in accuracy and mean average precision.Specifically,the Swin Transformer-YOLO model had a re-call rate of 77.6%,an mAP@0.5∶0.95 of 71%,and an F1 score of 0.84.In summary,this study provides a technical means for the ac-curate identification of corn leaf diseases in complex environments and offered new insights for small target detection.