基于改进YOLOv8n轻量化的番茄叶霉病发病程度分级检测

打开文本图片集

中图分类号：S126；TP391.4 文献标识码：A 文章编号： 1000-4440（2025）10-1985-12

Abstract：To further improve the accuracyand efficiency of tomato leaf mold severity grading recognition，reduce the complexityandweightof thedetectionmodel，andfacilitatedeploymentonmobiledevices，thisstudyimproved the YOLO v8n modeland proposeda lightweight disease severitygrading detection method.The UIB module from MobileNetV4 wasintroduced toreplacetheC2fconvolutional layerintheYOLOv8nmodel，reducing computationalloadandthenumber of parameters to metthe requirements forlightweight deployment on mobile devices.Thecascaded group attntion（CGA） modulewasincorporatedintothehighest-dimensionallayerofthebackbonenetwork，alongwiththeintroductionofpositionalbias.Aditionally，theoriginal decoupleddetectionheadwasreplacedwithadualatention-enhanceddetectionhead. Thesemodifications enabled the model toachieve precise localizationof leaf mold symptoms.Theresults demonstrated that thecascaded group attention（CGA）module yielded the mostsignificant improvement in model performance. Compared to the YOLO v8n model，the YOLO v8n-UCDAEmodel achieved increases of 2.O percentage points in P ，7.7percentage pointsin R ，3.8percentage points in mAP50 ，and 2.9 percentage points in mAP50-95 . Meanwhile， the computational load and the number ofparameterswere reduced by 43.33% and 32.86% ，respectively.Compared withothermainstream models，the YOLOv8n-UC-DAEmodeldeveloped inthis studyiscapableof meeting therequirementsforgrading theseverityoftomatoleaf moldandhasefectivelyaddressed thechallengesassociatedwithmobiledeployment.

Key words： YOLO v8n model； tomato leaf mold；severity grading

番茄叶霉病是一种真菌病害，其病原菌是黄枝孢菌（Cladosporiumfulvum）[1]，这种病害在潮湿、温暖的环境中容易发生，会严重影响番茄产量和品质。（剩余14151字）