基于改进YOLOv8的轻量化水稻病虫害识别模型研究

打开文本图片集

中图分类号：S126，TP391.41 文献标识码：A

文章编号：0439-8114（2025）08-0010-07

DOI：10.14088/j.cnki.issn0439-8114.2025.08.002 开放科学（资源服务）标识码（OSID）：口

Research on a lightweight rice pests and diseases recognition model based on the improved YOLOv8

LI Peng-fei， ZENG Jing

Economics and Management School，Yangtze University，Jingzhou 434O23，Hubei，China）

Abstract：BasedontheYOLOv8 model，the ShufleNetv2 moduleandtheConv_MaxPool modulewere introducedsimultaneouslyto construct theimprovedYOLOv8 model（YOLOv8-ShuffleNetv2-Conv_MaxPool）.Byintegrating the ShuffleNetv2 moduleand the Conv - MaxPol moduleinto the YOLOv8 model，the improved YOLOv8 modelsignificantly enhanced thecomprehensiveperformance ofricepestsanddiseasesdetectionhilemaintainingitslightweightdsign，efectivelyreducingboththefalsedetectionateandthe miseddetectionrate.TheimprovedYOLOv8modeldemonstrated excelentperformanceacrossmultipledatasets，furthervalidating itsrobustnessandgeneralizationabilityAblationstudiesdemonstratedthat，onthcustomdataset，comparedtotheoriginalYOOv8 model，theimproedYOLOv8odelacievedincreasesof3.73percentagepointsinaccuracy3.56percentagepointsiprision 3.78 percentagepointsincall，and3.73percentagepoitsinF-score，hilemaintainingaparametersizeofoly24.80MB.Onthe Coco128 dataset，the improved YOLOv8 model performed the best，with allkey metrics averaging approximately 88.00% ，significantlyoutperformingtheoriginalYOLOv8model，theYOLOv8-ShufleNetv2model，andtheYOLOv8-Con_MaxPolmodel.This model effectively enabled rapid and accurate recognition of rice pests and diseases in practical production environments.

Key words：rice pests and diseases；improved YOLOv8 model; lightweight design; recognition model

水稻是全球重要的粮食作物，其产量直接关系到全球粮食安全。（剩余6713字）