烟苗整齐度评估分析模型研究

打开文本图片集

（1.，南宁53000；2.，重庆400715；3.，重庆奉节404600)

中图分类号：S572；S126文献标识码：A

文章编号：1007-5119（2025）02-0101-12

An Analytical Model for Assessing Tobacco Seedling Uniformity

LI Qunling1, SUN Jiazhao, WANG Zhenguo³, CHENG ,RAN Yu'ao², LI ,DING Wei1*

(1.ChinaTobaoagxistrialCoLtd.,ago,Cina;2.CllgeoantrotectioouthwstUiy

Chongqing 40o715,China;3.FengjieBranchofChongqingTobacoCompany,ChinaNationalTobacco Corporatio,Fengie 404600, Chongqing, China)

Abstract:Toacheverapidassessmentandeficentanalysisoftheuniforityoftobaccosedlings inintensiveedlingfactorythis studyemploysageneralizedaditive model(GAM)toanalyetobaccoseed nurserydataadscrnfor idicatorsoftobaccoseedling uniformity. We evaluatedRandomForest algorithm,BPNeuralNetworkalgorithm,and Support VectorMachine(SVM)algorithm. Particle Swarm Optimization (PSO)isthen appliedtooptimizeeachof these modelsseparately.Thisstudyconstructed image recognitionmodelsforassessingtheuniforityof tobaccseednurseryusingdeepleaingalgorithms,twooptimizers,Adamand Nadam,specificallyAlexNet,ResNet-10l,and GoogleNet.Teresearchresultsindicated thattheplantheight,stemcicuferee, and numberofefectiveleaves of tobaccoseedlings hadasignificantimpactontheuniformityofthe tobaccoseedlings.ThePartcle Swarm Optimized Random Forest model demonstrates the best performance, with accuracy of 8 8 . 0 0 % valueof O.69,andMean AbsoluteError(MAE）of 0.13.TheAdam-GoogLeNet modelshows thebestrecognition performance,averagingacuracyof 9 3 . 8 9 % Overall,findings of thisstudy provideareasonable basis for thescientificevaluationoftobacco nurserybeduniformityand offer support for the development of tobacco seedling uniformity image recognition systems.

Keywords: tobacco seedbed; uniformity index; deep learning; uniformity model; image recognition

烟草是一种重要的经济作物，烟苗整齐度是烟草生产的关键指标之一，直接影响田间管理效果。（剩余23225字）