基于YOLOv8n改进的植物病害检测算法

打开文本图片集

中图分类号：TP183 文献标识码：A文章编号：1006-8228(2025)09-01-06

Plant Disease Detection Algorithm Based on Improved YOLOv8n

Fu Yanjun,Guo Min

(CollgefllUs

Abstract:Aimingattheproblemsofsimilartexturefeaturesofdiferentdiseasesandseriousocclusioncoverageoftargetsinplant diseasedetectionundercomplexbackground,aplantdiseasedetectionmethodnamedSPR-YOLOv8basedonimprovedYOLOv8n wasproposed.FirstlytheSPDConvmoduleisintroducedtoenhancetheabilityoffine-grainedfeatureextraction.Secondlythe PAmechanismisaddedinthebackbonenetworktooptimizethefocusareaofthemodelandimprovethedetectionacuracy. Final,thedetectionheadisreplacedbytheRT-DETRdecodertoenancethefeatureunderstandingabilityofthemodel.The experimental results show that compared with the base model, mAP@50 and mAP@50:95 increased by 3.1 and 2.6 percentage points respectively,which proved the effectiveness of the improved method.

Keywords:YOLOv8n;Plant DiseaseDetection;Fine-Grained FeatureExtraction;AttentionMechanism;RT-DETR

0引言

植物病害对农作物产量有着极大的影响，若早期能及时有效地发现植物病害，就可针对性地进行消杀安排，保障作物产量，对于粮食安全有着至关重要的意义。（剩余10999字）