基于无人机多光谱的烟苗计数模型研究

打开文本图片集

关键词种植密度；无人机多光谱；YOLOv5s;深度学习中图分类号S126文献标识码A文章编号 0517-6611（2025）17-0199-07doi： 10.3969/j. issn. 0517-6611. 2025. 17. 041

开放科学（资源服务）标识码（OSID）：

Research on Tobacco Seedling Counting Model Based on UAV Multispectral Imagery

ZHOUXi-xin，LUing-jian，HUZi-ming²etal（1.ColegeofBiologicalSciencesndTehnolgy，HunanAgriculuralUivesity Changsha，Hunan 410128;2.Baoshan Branch of Yunnan Tobacco Company，Baoshan，Yunnan 678000）

AbstractDurigthulioffucudtbaoeal-tacqsioofntigdsitysucialfoesugteielddaliyf tobaccoleaves.Tddssidlcaatsiatisinhraplatds 87asthetestsaple.ltispectraliageryncudingvisiblelight（GB）adDireceVeeationIdex（DV）ndd Soil-AdjustedVegetationIndex（OSAV）ascolectedusingUAVtechnogduringthetransplantingperiod.TheYOLOsectdetectio modelwasusedtoountthetobaccosedlingsanditspeformancewasomparedwihSSDandFaster-Rbjectdetectionodels.The resultsshowedtatteaageprecisiomA）ofteLOvsdelfrbacdigdeectionderthtesectraliasBI and OSAVI was 96.5% ，98. 6% ，and 94.7% ，respectively，with a detection speed of 58.3-58.8 frames per second（fps）.In comparison，the detection acuracyfSDandFaster-RCNNaslowerthanthatofYOLOvs.eNDVmulispecralimagerytraidwiththYOLO5sodelacevedhiger average precision and detection speed，making it more suitable for real-time counting of tobacco planting density.

Key wordsPlanting density; UAV multispectral imagery ; YOLOv5s;Deep learning

烤烟作为我国欠发达地区重要的经济作物之一，在区域经济发展中具有重要地位[1]。（剩余6553字）