[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-1818":3,"related-1818":85},{"id":4,"title":5,"url":6,"summary":7,"summary_zh":8,"content":8,"source_name":9,"source_url":8,"published_at":10,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":14,"score_detail":15,"sources":23,"tags":25,"primary_topic":31,"secondary_topics":37,"classification":44,"search_phrases":69,"slug":72,"view_count":73,"doi":8,"paper":74,"created_at":84},1818,"From Orbit to Leaf: A Critical Systematic Review of Artificial Intelligence for Multi-Scale Crop Monitoring and Soybean Leaf Disease Diagnosis（IJAIML 2026, 6(9s), 471-498）","https:\u002F\u002Fwww.svedbergopen.com\u002Findex.php\u002Fijaiml\u002Farticle\u002Fview\u002F1516","依据PRISMA 2020筛选4286篇记录，保留2019—2026年84项主要研究（41项轨道、17项航空、26项近距），以大豆叶病识别为近距测试案例。引入AGRI-CRITIC八维评估标准并综合为Field Readiness Index（FRI）。中位报告准确率93.9%但中位FRI仅37.5\u002F100，仅22.6%研究超过田间就绪阈值60，且报告准确率与评估集大小呈1.9%\u002F十倍样本负相关——小样本乐观的标志。识别七类反复失败模式，发布FIELD-READY十项报告协议与三阶段研究议程。",null,"International Journal of Artificial Intelligence and Machine Learning 2026, 6(9s)","2026-09-05T00:00:00Z","论文",10,false,81,{"impact":16,"substance":17,"depth":18,"authority":19,"freshness":20,"relevant":21,"comment":22},18,24,20,12,7,1,"系统综述揭示AI作物监测报告准确率与田间实用性差距，提出评估标准与报告协议，对智慧农业落地有重要参考价值。",[24],{"name":9,"url":6},[26,27,28,29,30],"农业人工智能","系统综述","作物监测","大豆叶病","田间就绪",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},"A4","soybean-oilseeds","大豆与油料","\u002Ftopic\u002Ffield-crops\u002Fsoybean-oilseeds","A",[38],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},"I2","agricultural-ai","农业AI与大模型","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-ai","I",{"status":45,"locked":13,"version":46,"method":47,"evidence":48},"classified","2026-10-22.2","rules",{"candidates":49,"agricultural_evidence":52},[50,58,64],{"code":32,"score":51,"supported":52,"title":53,"abstract":55,"tags":57},11,true,[54],"soybean",[56],"大豆",[],{"code":39,"score":59,"supported":52,"title":60,"abstract":62,"tags":63},9,[61],"artificial intelligence",[],[26],{"code":65,"score":21,"supported":13,"title":66,"abstract":67,"tags":68},"I3",[],[],[28],[70,71],"农业人工智能 作物监测 大豆叶病 田间就绪","农业人工智能 作物监测","农业人工智能作物监测大豆叶病田间就绪-1818",2,{"doi":8,"openalex_id":8,"authors":75,"venue":8,"cited_by_count":76,"oa_url":8,"card":77,"direction":81,"ingested_from":83},[],0,{"tldr":78,"method":79,"finding":80,"direction":81,"opportunity":82},"系统综述AI用于多尺度作物监测与大豆叶病诊断，提出FRI评估标准与报告协议。","PRISMA筛选84项研究，提出AGRI-CRITIC八维评估与FRI指数。","报告准确率中位93.9%但FRI仅37.5，小样本导致乐观偏差，仅22.6%达到田间就绪。","农业人工智能与决策模型","可开发考虑样本量与田间条件的模型评估方法，或构建跨尺度数据集以提升实际部署可靠性。","agent","2026-09-07T00:04:41.715297Z",{"total":86,"page":21,"page_size":86,"items":87},6,[88,167,229,317,367,441],{"id":89,"title":90,"url":91,"summary":92,"summary_zh":93,"content":8,"source_name":94,"source_url":91,"published_at":95,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":96,"score_detail":97,"sources":101,"tags":103,"primary_topic":108,"secondary_topics":109,"classification":111,"search_phrases":132,"slug":135,"view_count":73,"doi":136,"paper":137,"created_at":166},3986,"YOLOv8-MST: a lightweight and efficient real-time detection and counting deep learning algorithm for double-grain peanut seeding","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.compag.2026.112481","YOLOv8-MST: a lightweight and efficient real-time detection and counting deep learning algorithm for double-grain peanut seeding。Computers and Electronics in Agriculture","YOLOv8-MST：一种用于双粒花生播种的轻量高效实时检测与计数深度学习算法。《计算机与电子农业》","Computers and Electronics in Agriculture","2026-10-01T00:00:00Z",72,{"impact":19,"substance":18,"depth":98,"authority":99,"freshness":59,"relevant":21,"comment":100},17,14,"核心期刊发表的轻量化实时检测计数算法，对花生精量播种智能化有实用价值，但属细分技术进展，影响面有限。",[102],{"name":94,"url":91},[104,26,105,106,107],"智慧农业","目标检测","精准农业","花生播种",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[110],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},{"status":45,"locked":13,"version":46,"method":47,"evidence":112},{"candidates":113,"agricultural_evidence":52},[114,121,127],{"code":39,"score":115,"supported":52,"title":116,"abstract":118,"tags":120},16,[117],"deep learning",[119,117],"深度学习",[26,105],{"code":32,"score":99,"supported":52,"title":122,"abstract":124,"tags":126},[123],"peanut",[125,123],"花生",[],{"code":128,"score":73,"supported":13,"title":129,"abstract":130,"tags":131},"I1",[],[],[104,106],[133,134],"YOLOv8-MST 花生播种","双粒花生 实时检测计数","YOLOv8-MST花生播种-3986","10.1016\u002Fj.compag.2026.112481",{"doi":136,"openalex_id":138,"authors":139,"venue":94,"cited_by_count":76,"oa_url":8,"card":160,"direction":81,"ingested_from":165},"W7215133023",[140,143,145,147,149,151,153,155,158],{"name":141,"orcid":142},"Baogang Li","https:\u002F\u002Forcid.org\u002F0000-0002-3430-7508",{"name":144,"orcid":8},"Enxi Liu",{"name":146,"orcid":8},"Yijia Chen",{"name":148,"orcid":8},"Zhipeng Feng",{"name":150,"orcid":8},"Yuhuan Li",{"name":152,"orcid":8},"Zunmin Liu",{"name":154,"orcid":8},"Jingchun Tang",{"name":156,"orcid":157},"Jiang Liu","https:\u002F\u002Forcid.org\u002F0000-0001-6746-9987",{"name":159,"orcid":8},"Fazhan Yang",{"tldr":161,"method":162,"finding":163,"direction":81,"opportunity":164},"提出YOLOv8-MST轻量算法，实现双粒花生播种实时检测与计数。","基于YOLOv8改进的轻量深度学习模型，用于花生播种图像检测计数。","YOLOv8-MST在保持高精度的同时实现轻量化实时检测与计数。","可探索轻量模型在多种作物播种质量实时监测中的泛化与边缘部署。","openalex","2026-10-02T23:30:03.835851Z",{"id":168,"title":169,"url":170,"summary":171,"summary_zh":8,"content":8,"source_name":172,"source_url":8,"published_at":173,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":174,"score_detail":175,"sources":179,"tags":181,"primary_topic":185,"secondary_topics":186,"classification":188,"search_phrases":216,"slug":219,"view_count":207,"doi":220,"paper":221,"created_at":228},2859,"县域大豆产量预测的深度学习方法:利用大规模环境数据——Frontiers in Artificial Intelligence","https:\u002F\u002Fwww.frontiersin.org\u002Fjournals\u002Fartificial-intelligence\u002Farticles\u002F10.3389\u002Ffrai.2026.1914697\u002Ffull","论文提出了基于深度学习的大规模环境数据县域大豆产量预测方法。利用县级大豆产量数据和环境变量构建预测模型,评估多种深度学习架构(卷积神经网络、循环神经网络、Transformer等)在大豆产量预测中的表现,并与传统统计回归模型进行比较。结果显示,深度学习方法在预测精度和稳定性方面优于传统模型,能够更好地捕捉环境因素与产量之间的非线性关系,为农业政策制定和粮食安全评估提供数据支持。","Frontiers in Artificial Intelligence","2026-09-17T00:00:00Z",77,{"impact":115,"substance":176,"depth":98,"authority":177,"freshness":12,"relevant":21,"comment":178},21,13,"论文以县级大规模环境数据对比多种深度学习架构与传统回归模型，方法新颖、结论可靠，对农业政策与粮食安全评估有参考价值，值得进入每日精选。",[180],{"name":172,"url":170},[104,26,182,183,184],"粮食安全","大豆产量预测","环境数据",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[187],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},{"status":45,"locked":13,"version":46,"method":47,"evidence":189},{"candidates":190,"agricultural_evidence":52},[191,195,199,205,212],{"code":39,"score":18,"supported":52,"title":192,"abstract":193,"tags":194},[119,61],[119],[26],{"code":32,"score":51,"supported":52,"title":196,"abstract":197,"tags":198},[56],[56],[],{"code":200,"score":201,"supported":52,"title":202,"abstract":203,"tags":204},"A8",4,[],[182],[182],{"code":206,"score":207,"supported":52,"title":208,"abstract":209,"tags":211},"J1",3,[],[210],"农业政策",[],{"code":128,"score":21,"supported":13,"title":213,"abstract":214,"tags":215},[],[],[104],[217,218],"农业人工智能 大豆产量预测 智慧农业 环境数据","农业人工智能 大豆产量预测","农业人工智能大豆产量预测智慧农业环境数据-2859","10.3389\u002Ffrai.2026.1914697\u002Ffull",{"doi":220,"openalex_id":8,"authors":222,"venue":8,"cited_by_count":76,"oa_url":8,"card":223,"direction":81,"ingested_from":83},[],{"tldr":224,"method":225,"finding":226,"direction":81,"opportunity":227},"用深度学习结合大规模环境数据预测县域大豆产量，并对比多种网络架构与传统模型。","县级大豆产量与环境变量数据，采用CNN、RNN、Transformer等深度学习","深度学习在预测精度和稳定性上优于传统统计回归，能更好捕捉非线性关系。","可探索多模态环境数据融合与可解释性，提升跨区域迁移和极端气候下的预测鲁棒性。","2026-09-18T00:03:31.241582Z",{"id":230,"title":231,"url":232,"summary":233,"summary_zh":234,"content":8,"source_name":235,"source_url":232,"published_at":236,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":237,"score_detail":238,"sources":242,"tags":244,"primary_topic":248,"secondary_topics":249,"classification":255,"search_phrases":278,"slug":281,"view_count":201,"doi":282,"paper":283,"created_at":316},2320,"Genotype-Aware Prediction of Soybean Seed Composition from Multimodal UAV Imagery of the Standing Crop","https:\u002F\u002Fdoi.org\u002F10.3390\u002Frs18183121","Geospatial artificial intelligence (GeoAI) integrates multimodal remote sensing with deep learning to model complex agricultural systems at scale. Within this framework, accurate and non-destructive prediction of seed composition from in-season standing crops is essential for breeding and precision agriculture. This study developed an end-to-end convolutional neural network (CNN) framework to estimate eight seed traits (protein, oil, sucrose, fiber, starch, ash, complex and simple carbohydrates) from UAV-based multisensor imagery and associated genotype and phenological metadata. A total of 372 soybean samples were collected over two growing seasons (2020–2021) from two fields in Missouri, with UAV flights capturing multispectral (MSI), thermal (THR), and LiDAR (LDR) data at four time points spanning vegetative to reproductive growth stages. CNN models were trained in single- and multi-date configurations, incorporating genotype (GEN) and days after sowing (DAS) as additional features. The highest accuracy was achieved for sucrose (R2 = 0.80), followed by simple carbohydrate (R2 = 0.70) and starch (R2 = 0.55), with notable gains from GEN and DAS. Multi-date models incorporating earlier acquisitions often matched or outperformed later or all-date combinations. Among modalities, MSI provided the most robust estimates, with limited added value from LDR or THR. Unlike feature-based pipelines prone to multicollinearity, this image-to-trait approach enables automated, scalable prediction of soybean seed composition for in-season, field-level assessment.","地理空间人工智能（GeoAI）将多模态遥感与深度学习相结合，以在大尺度上对复杂农业系统进行建模。在该框架下，对当季未收获作物进行准确且无损的种子成分预测，对于育种和精准农业至关重要。本研究开发了一种端到端的卷积神经网络（CNN）框架，利用基于无人机（UAV）的多传感器影像以及相关的基因型和物候元数据，估算八种种子性状（蛋白质、油分、蔗糖、纤维、淀粉、灰分、复杂碳水化合物和简单碳水化合物）。研究在2020—2021两个生长季内，从密苏里州两块田地共采集372份大豆样本，无人机飞行在从营养生长期到生殖生长期的四个时间点获取了多光谱（MSI）、热红外（THR）和激光雷达（LDR）数据。CNN模型以单日期和多日期配置进行训练，并将基因型（GEN）和播种后天数（DAS）作为附加特征纳入。蔗糖的估算精度最高（R² = 0.80），其次为简单碳水化合物（R² = 0.70）和淀粉（R² = 0.55），且GEN和DAS的加入带来了显著提升。纳入早期采集数据的多日期模型往往与后期或全日期组合的表现相当，甚至更优。在各模态中，MSI提供了最稳健的估算结果，而LDR或THR的附加价值有限。与易受多重共线性影响的特征工程流程不同，这种从图像到性状的方法能够实现大豆种子成分的自动化、可扩展预测，用于当季田块尺度的评估。","Remote Sensing","2026-09-11T00:00:00Z",80,{"impact":16,"substance":239,"depth":16,"authority":99,"freshness":240,"relevant":21,"comment":241},22,8,"基于多模态无人机影像与基因型信息的大豆籽粒成分无损预测研究，方法新颖、数据扎实，对智慧育种与精准农业有参考价值。",[243],{"name":235,"url":232},[104,245,26,246,247],"无人机","遥感","大豆育种",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[250,251],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},{"code":65,"slug":252,"name":253,"path":254,"parent_code":43},"remote-sensing","遥感与农情监测","\u002Ftopic\u002Fdigital-agriculture\u002Fremote-sensing",{"status":45,"locked":13,"version":46,"method":47,"evidence":256},{"candidates":257,"agricultural_evidence":52},[258,263,267,273],{"code":32,"score":259,"supported":52,"title":260,"abstract":261,"tags":262},15,[54],[56,54],[247],{"code":39,"score":12,"supported":52,"title":264,"abstract":265,"tags":266},[],[119,117,61],[26],{"code":65,"score":12,"supported":52,"title":268,"abstract":269,"tags":272},[],[246,270,271],"remote sensing","multispectral",[246],{"code":128,"score":20,"supported":52,"title":274,"abstract":275,"tags":277},[],[106,276],"precision agriculture",[104],[279,280],"农业人工智能 大豆育种 智慧农业 无人机","农业人工智能 大豆育种","农业人工智能大豆育种智慧农业无人机-2320","10.3390\u002Frs18183121",{"doi":282,"openalex_id":284,"authors":285,"venue":235,"cited_by_count":76,"oa_url":232,"card":310,"direction":314,"ingested_from":165},"W7212223520",[286,289,292,295,298,300,302,305,308],{"name":287,"orcid":288},"Vasit Sagan","https:\u002F\u002Forcid.org\u002F0000-0003-4375-2096",{"name":290,"orcid":291},"Kristen Rhodes","https:\u002F\u002Forcid.org\u002F0000-0002-8469-5043",{"name":293,"orcid":294},"Sourav Bhadra","https:\u002F\u002Forcid.org\u002F0000-0002-5832-4695",{"name":296,"orcid":297},"Haireti Alifu","https:\u002F\u002Forcid.org\u002F0000-0002-7369-6657",{"name":299,"orcid":8},"Aviskar Giri",{"name":301,"orcid":8},"Ashutosh Pawar",{"name":303,"orcid":304},"Bishal Roy","https:\u002F\u002Forcid.org\u002F0000-0001-9912-5505",{"name":306,"orcid":307},"Supria Sarkar","https:\u002F\u002Forcid.org\u002F0000-0001-9467-5786",{"name":309,"orcid":8},"Felix Fritschi",{"tldr":311,"method":312,"finding":313,"direction":314,"opportunity":315},"用无人机多模态影像与基因型数据训练CNN，预测大豆八种种子成分。","372份大豆样本，多光谱\u002F热红外\u002FLiDAR影像加基因型与播后天数，端到端CNN","蔗糖预测最优(R²=0.80)，基因型与播期提升精度，多光谱最稳健。","农业遥感与作物表型","可探索基因型-表型-环境交互建模，并迁移至其他作物与多时相早期预测。","2026-09-13T23:30:22.776177Z",{"id":318,"title":319,"url":320,"summary":321,"summary_zh":8,"content":322,"source_name":323,"source_url":8,"published_at":324,"category":325,"cover_url":8,"hotness":12,"is_selected":13,"score":326,"score_detail":327,"sources":330,"tags":332,"primary_topic":336,"secondary_topics":337,"classification":339,"search_phrases":362,"slug":365,"view_count":73,"doi":8,"paper":8,"created_at":366},2100,"安徽农业大学发布大豆垂直大模型\"丰菽\"2.0 构建我国首个大豆全生命周期生成式AI育种平台","https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ4OTEwNTQ.html","在第三十二届全国大豆科研生产研讨会上,安徽农业大学大豆智能设计育种创新团队联合中国农业科学院作物科学研究所共同研发的大豆领域垂直大模型\"丰菽\"2.0正式发布。该模型集成豆百科、豆分子、豆文献、豆病害、豆表型、豆育种六大功能模块,汇聚大豆核心种质资源、基因组、表型影像、科研文本等多源数据,构建\"多源验证—智能融合—可信输出\"的知识技术架构。科研人员用手机拍摄大豆叶片即可进行病害诊断与表型特征提取,实现\"随拍随诊\"。","在东港\n\n官方移动发布平台\n\n[](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ4OTEwNTQ.html)\n\n![Image 1](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fservice2\u002Farticle\u002Farticle-detail-share\u002Fimages\u002Ficon-voice.png)听新闻\n\n[Audio 2](https:\u002F\u002Fstream-spk.iqilu.com\u002F349\u002F2026\u002F09\u002F07\u002F44891054.mp3?1788887065172)\n\n大豆垂直大模型“丰菽”2.0正式发布\n\n00:00\n\n# 大豆垂直大模型“丰菽”2.0正式发布\n\n人民日报\n\n2026-09-07 08:48\n\n记者从安徽农业大学获悉：在第三十二届全国大豆科研生产研讨会上，由该校大豆智能设计育种创新团队联合中国农业科学院作物科学研究所共同研发的大豆领域垂直大模型“丰菽”2.0正式发布。\n\n该模型构建了我国首个面向大豆全生命周期的生成式人工智能育种平台，有望推动大豆育种从依赖经验向数据辅助决策转变，为提升大豆单产提供新的技术路径。\n\n据介绍，“丰菽”2.0集成了豆百科、豆分子、豆文献、豆病害、豆表型、豆育种六大功能模块，汇聚大豆核心种质资源、基因组、表型影像、科研文本等多源数据，构建起“多源验证—智能融合—可信输出”的知识技术架构，面对同一个大豆专业问题，不同模型可分别进行独立分析，再由“丰菽”总结模块结合领域知识、私有数据和专家规则，对回答中的共识、差异及证据完整性进行结构化比对与综合研判，有效降低了单一模型的回答偏差，输出更加清晰可靠的专业结论。\n\n经过优化，科研人员只需用手机拍摄大豆叶片或植株，即可依托大模型进行病害诊断与表型特征提取，实现“随拍随诊”；同时打通了电脑端与移动端的数据壁垒，育种试验记录、分子标记辅助育种数据及候选基因功能分析结果可实现云端实时同步，方便科研团队随时随地开展虚拟组合设计与数据查阅。\n\n目前，“丰菽”2.0可围绕大豆重要农艺性状，辅助开展病害诊断、育种亲本筛选、虚拟组合设计、表型数据解析、分子标记辅助育种和候选基因功能分析等核心场景应用，能够显著提高科研数据处理效率，减少重复性工作，实现优异等位基因智能挖掘和最优亲本组合虚拟设计。\n\n编辑：许艺璇 王倩 费洪涛 王西辉\n\n版权说明\n\n特别声明：本文为“在东港”客户端“海曲号”作者上传并发布，仅代表作者观点。“在东港”客户端仅提供信息发布平台。\n\n[打开“在东港”看评论 ![Image 2](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fservice2\u002Farticle\u002Farticle-detail-share\u002Fimages\u002Ficon_comment.png) ![Image 3](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fservice2\u002Farticle\u002Farticle-detail-share\u002Fimages\u002Ficon-coll.png) ![Image 4](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fservice2\u002Farticle\u002Farticle-detail-share\u002Fimages\u002Ficon_parse.png)](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ4OTEwNTQ.html)\n\n## 相关资讯\n\n### [我国已累计发布8批121个新职业 数字职业、绿色职业均超百个 15小时前 观天下 ![Image 5](https:\u002F\u002Fimg12.litenews.cn\u002F10293\u002Fsucaiku\u002Fcompress\u002F202312\u002F06\u002F073864c224ae461eb225b4dfb984e096.png)](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ5MzYzNjA.html)### [从机器人“教练”到适老化改造设计师，11个新职业23个新工种发布 新需求一冒头 新职业就“破土” 15小时前 观天下 ![Image 6](https:\u002F\u002Fimg12.litenews.cn\u002F10293\u002Fsucaiku\u002Fcompress\u002F202602\u002F10\u002F32831a81a0094f3b89fb9f6d3bc1e1dc.png)](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ5MzYzNTk.html)### [一车一主题、一厢一景致 “鲁甘协作”冠名列车首发 15小时前 观天下 ![Image 7](https:\u002F\u002Fimg12.litenews.cn\u002F10293\u002Fsucaiku\u002Fcompress\u002F202602\u002F10\u002F32831a81a0094f3b89fb9f6d3bc1e1dc.png)](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ5MzYzNTg.html)### [传承弘扬英烈精神 筑牢国家安全屏障”主题活动举办 15小时前 观天下 ![Image 8](https:\u002F\u002Fimg12.litenews.cn\u002F10293\u002Fsucaiku\u002Fcompress\u002F202602\u002F10\u002F32831a81a0094f3b89fb9f6d3bc1e1dc.png)](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ5MzYzNTQ.html)### [潍坊推动医保、医疗、医药数据融合 人均住院费用下降一千八百余元 15小时前 观天下 ![Image 9](https:\u002F\u002Fimg12.litenews.cn\u002F10293\u002Fsucaiku\u002Fcompress\u002F202602\u002F10\u002F32831a81a0094f3b89fb9f6d3bc1e1dc.png)](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ5MzYzNTA.html)### [从被看见到被照亮山东成就141万“筑梦人” 15小时前 观天下 ![Image 10](https:\u002F\u002Fimg12.litenews.cn\u002F10293\u002Fsucaiku\u002Fcompress\u002F202602\u002F10\u002Fef0b650c0ffe4013becd762b6a1dc423.png)](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ5MzYzNDg.html)### [三尺讲台，四季耕耘——“齐鲁最美教师”用行动和青春诠释“何以为师” 15小时前 观天下 ![Image 11](https:\u002F\u002Fimg12.litenews.cn\u002F10293\u002Fsucaiku\u002Fcompress\u002F202602\u002F10\u002F32831a81a0094f3b89fb9f6d3bc1e1dc.png)](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ5MzYzNDY.html)### [前8个月山东“新三样”出口增长27.7% 15小时前 观天下 ![Image 12](https:\u002F\u002Fimg12.litenews.cn\u002F10293\u002Fsucaiku\u002Fcompress\u002F202602\u002F10\u002F32831a81a0094f3b89fb9f6d3bc1e1dc.png)](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ5MzYzMzg.html)### [95家鲁企亮相2026服贸会 15小时前 观天下 ![Image 13](https:\u002F\u002Fimg12.litenews.cn\u002F10293\u002Farticle\u002F202609\u002F10\u002F716cf80314ad43ce8c5ddb33954e11e1.jpg)](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ5MzYzMjE.html)### [与法同行 护航成长 15小时前 观天下 ![Image 14](https:\u002F\u002Fimg12.litenews.cn\u002F10293\u002Farticle\u002F202609\u002F10\u002Fddda5e025d104b92befddaefd6b221de.jpg)](https:\u002F\u002Fiapp.rzdgrmt.cn\u002Fshare\u002FYS0zNDktNDQ5MzYzMTY.html)","在东港\u002F今日头条","2026-09-07T00:00:00Z","报道",86,{"impact":328,"substance":239,"depth":98,"authority":177,"freshness":240,"relevant":21,"comment":329},26,"我国首个大豆全生命周期生成式AI育种平台发布，属种业与农业AI交叉领域的标志性突破，技术路径与功能模块描述具体，值得进入每日精选。",[331],{"name":323,"url":320},[26,333,334,56,335],"种业振兴","智能育种","垂直大模型",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[338],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},{"status":45,"locked":13,"version":46,"method":47,"evidence":340},{"candidates":341,"agricultural_evidence":52},[342,346,351,357],{"code":32,"score":19,"supported":52,"title":343,"abstract":344,"tags":345},[56],[56],[56],{"code":39,"score":19,"supported":52,"title":347,"abstract":349,"tags":350},[348],"大模型",[348],[26],{"code":352,"score":201,"supported":52,"title":353,"abstract":354,"tags":356},"F1",[],[355],"种质资源",[333],{"code":358,"score":21,"supported":13,"title":359,"abstract":360,"tags":361},"F2",[],[],[334],[363,364],"农业人工智能 垂直大模型 智能育种 种业振兴","农业人工智能 垂直大模型","农业人工智能垂直大模型智能育种种业振兴-2100","2026-09-11T00:04:21.173241Z",{"id":368,"title":369,"url":370,"summary":371,"summary_zh":372,"content":8,"source_name":373,"source_url":370,"published_at":374,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":375,"score_detail":376,"sources":378,"tags":380,"primary_topic":384,"secondary_topics":385,"classification":387,"search_phrases":407,"slug":410,"view_count":207,"doi":411,"paper":412,"created_at":440},3485,"Detection of Red Crown Rot of Soybean in Illinois Fields Using High-Resolution Satellite Imagery and Machine Learning","https:\u002F\u002Fdoi.org\u002F10.64898\u002F2026.09.22.753594","Red crown rot (RCR), caused by Calonectria ilicicola, is an emerging soybean disease in the U.S. Midwest for which scalable approaches to characterize within-field disease distribution are lacking. This study evaluated high-resolution PlanetScope satellite imagery for mapping RCR-affected soybean canopies across 15 commercial fields in Illinois surveyed during the 2024 and 2025 growing seasons. A total of 2,921 georeferenced canopy plots were classified as asymptomatic or RCR-affected and paired with six multispectral bands and seven vegetation indices. Spectral differences between classes were evaluated using linear mixed-effects models, and seven machine-learning classifiers representing linear, tree-based, neural-network, kernel, and probabilistic approaches were compared using spatially independent leave-one-field-out cross-validation. RCR-affected canopies exhibited increased reflectance in the visible and red-edge regions, reduced near-infrared reflectance, and lower vegetation-index values relative to asymptomatic canopies. All classifiers showed strong discrimination, with ROC-AUC values ranging from 0.963 to 0.982. Regularized logistic regression achieved the highest overall performance, with an accuracy of 0.945, balanced accuracy of 0.945, F1-score of 0.948, and ROC-AUC of 0.982 at the optimized decision threshold. Permutation analysis identified EVI, NDVI, and red reflectance as the most influential predictors across representative model architectures. Satellite-derived probability and classification maps generally corresponded with symptomatic canopy patterns observed in high-resolution UAV imagery, although mixed pixels reduced precision near disease-patch boundaries. These results demonstrate the potential of high-resolution satellite imagery for within-field mapping of RCR-associated canopy symptoms across independent commercial soybean fields.","由冬青丽赤壳菌（Calonectria ilicicola）引起的红冠腐病（RCR）是美国中西部一种新发大豆病害，目前尚缺乏可规模化表征田块内病害分布的方法。本研究评估了高分辨率PlanetScope卫星影像在2024年和2025年生长季对伊利诺伊州15块商业田块中受RCR影响的大豆冠层的制图能力。共有2,921个地理参考冠层样区被分类为无症状或受RCR影响，并与6个多光谱波段和7个植被指数配对。采用线性混合效应模型评估类别间的光谱差异，并通过空间独立的留一田块交叉验证比较了7种机器学习分类器，涵盖线性、基于树、神经网络、核函数和概率方法。与无症状冠层相比，受RCR影响的冠层在可见光和红边区域反射率增加，近红外反射率降低，植被指数值较低。所有分类器均表现出较强的判别能力，ROC-AUC值范围为0.963至0.982。正则化逻辑回归在优化决策阈值下取得了最高的整体性能，准确率为0.945，平衡准确率为0.945，F1分数为0.948，ROC-AUC为0.982。置换分析表明，在代表性模型架构中，EVI、NDVI和红光反射率是最具影响力的预测变量。卫星衍生的概率图和分类图总体上与高分辨率无人机影像中观察到的症状冠层模式一致，尽管混合像元降低了病害斑块边界附近的精度。这些结果证明了高分辨率卫星影像在独立商业大豆田块中对RCR相关冠层症状进行田块内制图的潜力。","bioRxiv (Cold Spring Harbor Laboratory)","2026-09-23T00:00:00Z",76,{"impact":115,"substance":239,"depth":16,"authority":19,"freshness":240,"relevant":21,"comment":377},"基于高分辨率卫星影像与机器学习实现大豆红冠腐病田间制图，方法扎实、数据规模可观，对作物病害遥感监测有参考价值。",[379],{"name":373,"url":370},[104,381,56,382,383],"机器学习","遥感监测","病害识别",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[386],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},{"status":45,"locked":13,"version":46,"method":47,"evidence":388},{"candidates":389,"agricultural_evidence":52},[390,394,399,403],{"code":32,"score":259,"supported":52,"title":391,"abstract":392,"tags":393},[54],[56,54],[56],{"code":39,"score":177,"supported":52,"title":395,"abstract":397,"tags":398},[396],"machine learning",[381],[381,383],{"code":65,"score":201,"supported":52,"title":400,"abstract":401,"tags":402},[],[271],[382],{"code":128,"score":21,"supported":13,"title":404,"abstract":405,"tags":406},[],[],[104],[408,409],"伊利诺伊 大豆 红冠腐病 卫星遥感","PlanetScope 大豆 病害 机器学习","伊利诺伊大豆红冠腐病卫星遥感-3485","10.64898\u002F2026.09.22.753594",{"doi":411,"openalex_id":413,"authors":414,"venue":373,"cited_by_count":76,"oa_url":434,"card":435,"direction":314,"ingested_from":165},"W7214126630",[415,418,421,423,425,428,431],{"name":416,"orcid":417},"Bruno Daniel Pugliese","https:\u002F\u002Forcid.org\u002F0009-0000-4795-4248",{"name":419,"orcid":420},"Juan Andrés Paredes","https:\u002F\u002Forcid.org\u002F0000-0002-3967-0197",{"name":422,"orcid":8},"Andres Fabian Ruiz",{"name":424,"orcid":8},"Norman Denis Bowman",{"name":426,"orcid":427},"Elhan S. Ersoz","https:\u002F\u002Forcid.org\u002F0000-0001-6930-6946",{"name":429,"orcid":430},"Nicolas Federico Martin","https:\u002F\u002Forcid.org\u002F0000-0002-1587-665X",{"name":432,"orcid":433},"Boris X. Camiletti","https:\u002F\u002Forcid.org\u002F0000-0002-1492-7988","https:\u002F\u002Fwww.biorxiv.org\u002Fcontent\u002Fbiorxiv\u002Fearly\u002F2026\u002F09\u002F23\u002F2026.09.22.753594.full.pdf",{"tldr":436,"method":437,"finding":438,"direction":314,"opportunity":439},"利用高分辨率卫星影像和机器学习在大豆田中检测红冠腐病。","PlanetScope多光谱影像、植被指数与7种机器学习分类器，留一田块交叉验证","各分类器ROC-AUC达0.963-0.982，正则化逻辑回归最优，EVI、NDVI和红波段最重要。","可探索多时相卫星影像与无人机融合，提升病害斑块边界混合像元识别精度。","2026-09-25T23:30:22.839600Z",{"id":442,"title":443,"url":444,"summary":445,"summary_zh":8,"content":446,"source_name":447,"source_url":8,"published_at":448,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":96,"score_detail":449,"sources":451,"tags":453,"primary_topic":456,"secondary_topics":457,"classification":459,"search_phrases":482,"slug":485,"view_count":486,"doi":8,"paper":8,"created_at":487},676,"UAV-Based Classification of Crop Phenological Stages Using Deep Learning(MDPI Drones 2026-08-17)","https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629","基于低空UAV RGB影像的作物物候期深度学习自动分类研究。利用11,489张向日葵、油菜、大豆、小麦和大麦图像,基于BBCH量表标注,采用预训练ResNet18通过迁移学习适应任务。两阶段训练:先冻结骨干优化分类头,再微调整个网络。油菜和大麦测试准确率100%,其他作物均超过99%,平均准确率达99.73%。结果支持低空UAV影像与深度学习在精准农业中选择性区域物候评估的潜力,但广泛部署需要跨独立田块、季节、地区和调查条件的验证。","Logical Operator Operator\n\nSearch Text\n\nSearch Type\n\n_add\\_circle\\_outline_\n\n_remove\\_circle\\_outline_\n\n[![Image 1: drones-logo](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fjournals\u002Fdrones-logo.png?7d8906730931ecdd)](https:\u002F\u002Fwww.mdpi.com\u002Fjournal\u002Fdrones)\n\n## Article Menu\n\nFont Type:\n\n_Arial_ _Georgia_ _Verdana_\n\nFont Size:\n\nAa Aa Aa\n\nLine Spacing:\n\n__ __ __\n\nColumn Width:\n\n__ __ __\n\nBackground:\n\nOpen Access Article\n\nby \nRavil I. Mukhamediev\n\n 1,2[](mailto:ravil.muhamedyev@gmail.com)[![Image 2: ORCID](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Forcid.png?0465bc3812adeb52?1786617101)](https:\u002F\u002Forcid.org\u002F0000-0002-3727-043X), \nValentin Smurygin\n\n 1[](mailto:valentinsmurygin@gmail.com)[![Image 3: ORCID](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Forcid.png?0465bc3812adeb52?1786617101)](https:\u002F\u002Forcid.org\u002F0009-0009-8699-159X), \nLiudmila Gorodetskaya\n\n 1[](mailto:ludmila260601@gmail.com)[![Image 4: ORCID](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Forcid.png?0465bc3812adeb52?1786617101)](https:\u002F\u002Forcid.org\u002F0009-0000-0524-8521), \nYan Kuchin\n\n 1,2[](mailto:ykuchin@mail.ru)[![Image 5: ORCID](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Forcid.png?0465bc3812adeb52?1786617101)](https:\u002F\u002Forcid.org\u002F0000-0002-5271-9071), \nAdilet Dauletuly\n\n 1[](mailto:adi.dauletuly@gmail.com)[![Image 6: ORCID](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Forcid.png?0465bc3812adeb52?1786617101)](https:\u002F\u002Forcid.org\u002F0009-0006-5290-5225), \nNursultan Kuldeyev\n\n 1[](mailto:kuldeevnursultan129@gmail.com)[![Image 7: ORCID](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Forcid.png?0465bc3812adeb52?1786617101)](https:\u002F\u002Forcid.org\u002F0009-0004-5906-1040), \nAdilkhan Symagulov\n\n 1,2,*[](mailto:asmogulove00@gmail.com) and \nIrina Fedorovich\n\n 3[](mailto:ifed@list.ru)\n\n1\n\nInstitute of Automation and Information Technology, Satbayev University (KazNRTU), Almaty 050013, Kazakhstan\n\n2\n\nInstitute of Information and Computational Technologies, Almaty 050010, Kazakhstan\n\n3\n\nAgriSoft LLP, Almaty 050060, Kazakhstan\n\n*\n\nAuthor to whom correspondence should be addressed.\n\nSubmission received: 4 June 2026 \u002F Revised: 5 August 2026 \u002F Accepted: 9 August 2026 \u002F Published: 17 August 2026\n\n## Highlights\n\n**What are the main findings?**\n\n*   A ResNet18 model based on transfer learning was developed to classify BBCH-defined phenological stages of five crops from low-altitude RGB imagery acquired by UAVs.\n\n*   Using 11,489 images of sunflower, rapeseed, soybean, wheat, and barley, the crop-specific models achieved high accuracy on internal test sets, with a mean accuracy of 99.73% under the studied low-altitude UAV imaging conditions.\n\n**What are the implications of the main findings?**\n\n*   The results support the potential of low-altitude UAV imagery and deep learning for localized phenological assessment of selected field zones in precision agriculture.\n\n*   BBCH-based labeling with both single stages and stage ranges can represent heterogeneous or transitional field conditions. However, broader deployment requires validation across independent fields, seasons, regions, and survey conditions.\n\n## Abstract\n\nThis study investigates the automatic classification of crop phenological stages from low-altitude UAV RGB imagery. The dataset included 11,489 images of five crops: sunflower, rapeseed, soybean, wheat, and barley. The images were annotated using the Biologische Bundesanstalt, Bundessortenamt und Chemische Industrie (BBCH) scale, with labels corresponding to either single stages or stage ranges to reflect heterogeneous field conditions and transitional crop states. A pretrained ResNet18 model was adapted to the task using transfer learning. Training was conducted in two stages: first, the classification head was optimized while the backbone remained frozen; second, the entire network was fine-tuned. The model achieved strong internal test accuracy across all crops, with 100% test accuracy for rapeseed and barley, more than 99% for the remaining crops, and a mean accuracy of 99.73% under the studied survey conditions. The results also compare favorably with previously reported studies on UAV-based phenological classification. Overall, the findings support the potential of low-altitude UAV imagery and deep learning for localized phenological assessment of selected field zones in precision agriculture, while broader deployment requires validation across independent fields, seasons, regions, and survey conditions.\n\n## 1. Introduction\n\nThe need to improve the sustainability and productivity of agricultural production in the face of climate change, resource scarcity, and increasing environmental demands is driving the transition from traditional technologies to precision agriculture (PA) [[1](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B1-drones-10-00629),[2](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B2-drones-10-00629)] and smart agriculture (SA) [[3](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B3-drones-10-00629)]. A key component of these technologies is monitoring, which provides regular spatial and temporal information on the condition of crops, soil, and environmental factors to support decision-making (from work planning to differentiated resource application and outcome forecasting). In the modern literature, the most common monitoring tasks include the following: (1) determining the phenological stages of crop development [[4](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B4-drones-10-00629)], (2) assessing the water status of plants and water stress to support irrigation decisions [[5](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B5-drones-10-00629)], (3) monitoring the nutrient (primarily nitrogen) status of crops to optimize fertilizers [[6](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B6-drones-10-00629)], (4) identifying diseases and pests [[7](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B7-drones-10-00629)], (5) monitoring weeds for targeted protection [[8](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B8-drones-10-00629)], (6) crop mapping [[9](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B9-drones-10-00629)] and forecasting yields based on observation data during the season [[10](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B10-drones-10-00629)], (7) monitoring soil condition, where along with the water regime, quality\u002Ffertility indicators (for example, organic carbon content as an integral indicator) are becoming increasingly important [[11](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B11-drones-10-00629)], and for arid regions—(8) soil salinization as a factor of degradation and reduced productivity [[12](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B12-drones-10-00629),[13](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B13-drones-10-00629)]. To address these challenges, modern agricultural monitoring utilizes remote, proximal, and ground-based sensing data, which differ primarily in the distance of the sensor from the observed object, spatial coverage, and level of detail. Remote sensing enables regular monitoring of large areas using satellite, aerial, and Unmanned Aerial Vehicle (UAV) platforms, while proximal and ground-based sensing focus on obtaining more detailed measurements near plants and from the field surface, making these approaches complementary in precision and intelligent farming systems [[14](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B14-drones-10-00629),[15](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B15-drones-10-00629)] ([Figure 1](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#fig_body_display_drones-10-00629-f001)).\n\n**Figure 1.** Key monitoring tasks in precision and smart agriculture using remote and proximal sensing.\n\nThis paper addresses the automatic classification of crop phenological stages using low-altitude UAV imagery, a task that is important for optimizing the timing of agronomic interventions and improving yield forecasting [[16](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B16-drones-10-00629)]. The proposed approach is intended for detailed phenological assessment of selected field zones identified during preliminary large-area monitoring using satellite imagery or higher-altitude UAV.\n\nTraditional methods for determining phenological phases, based on visual observations by experts, are labor-intensive, limiting their use for monitoring large areas. The development of remote sensing technologies, including the use of satellite data and images from unmanned aerial vehicles (UAVs), has significantly expanded the possibilities for collecting information on crop conditions [[4](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B4-drones-10-00629),[17](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B17-drones-10-00629)]. These data provide high spatial and temporal resolution and are widely used in agricultural monitoring.\n\nIn recent years, deep learning methods have played a key role in image analysis in agriculture. Reviews [[18](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B18-drones-10-00629),[19](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B19-drones-10-00629)] show that deep learning models demonstrate higher accuracy compared to traditional image processing methods. Convolutional neural networks (CNNs) are widely used for plant image classification and remote crop monitoring [[20](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B20-drones-10-00629),[21](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B21-drones-10-00629),[22](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B22-drones-10-00629)]. Pre-trained architectures such as ResNet [[23](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B23-drones-10-00629)] and detection models of the YOLO family, which have been successfully applied in agricultural tasks [[24](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B24-drones-10-00629)], can achieve high accuracy even with limited data.\n\nDespite the progress made, automatic classification of phenological stages remains a challenging task in practice. This is due to the high variability in plant appearance, the gradual nature of the transition between stages, and the limited availability of detailed labeled data. Crop heterogeneity further complicates the task: within a single field, plants at different stages of development may be present simultaneously.\n\nIn real-world settings, image labeling is often performed at the field level and reflects a mixture of phenological states. Therefore, this study uses an approach in which the label describes the dominant state of the plot and can correspond to either a single stage or a range of stages on the Biologische Bundesanstalt, Bundessortenamt und Chemische Industrie (BBCH) scale. This allows transitional field states to be represented explicitly during annotation.\n\nAccordingly, this study investigates the use of deep learning for automatic classification of the phenological stages of five crops—sunflower, rapeseed, soybean, wheat, and barley—from UAV images labeled according to the BBCH scale, including both single stages and stage ranges. To achieve this goal, the following tasks were completed: dataset formation and labeling, image preprocessing, model development and training based on the ResNet18 architecture, classification quality assessment, and analysis of the obtained results.\n\nThe paper is organized into six sections. [Section 2](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#sec2-drones-10-00629), Related Work, provides a focused overview of previous studies relevant to phenological classification of agricultural crops. [Section 3](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#sec3-drones-10-00629) briefly describes the study area, initial data, and methods. [Section 4](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#sec4-drones-10-00629) presents the experimental results, [Section 5](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#sec5-drones-10-00629) discusses the results and limitations, and [Section 6](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#sec6-drones-10-00629) summarizes the work and outlines directions for further research.\n\n## 2. Related Work\n\nPrevious studies have addressed crop phenological classification using manual field observations, semi-automated proximal sensing, satellite time series, and image-based approaches, including UAV and ground-based imagery [[14](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B14-drones-10-00629),[16](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B16-drones-10-00629),[25](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B25-drones-10-00629),[26](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B26-drones-10-00629),[27](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B27-drones-10-00629)]. These studies demonstrate that phenological information can be derived from different sensing platforms and modeling strategies, but performance and transferability depend strongly on data type, spatial and temporal resolution, labeling strategy, and validation design.\n\n### 2.1. Methods of Phenological Classification\n\nFor clarity, phenological classification methods can be grouped by the degree of automation and the type of data used ([Figure 2](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#fig_body_display_drones-10-00629-f002)). Manual methods include visual observations by experts in the field, the use of phenological diaries and scales, particularly the BBCH scale, as well as protocols for selective crop sampling [[25](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B25-drones-10-00629)].\n\n**Figure 2.** Classification of methods for determining crop phenological stages: manual, semi-automated, and automated approaches.\n\nManual observations are biologically interpretable and can capture transitional states, but they are labor-intensive, subjective, and difficult to scale. Semi-automated methods combine visual observations with proximal or handheld measurements and determine the stage using selected features or rules [[14](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B14-drones-10-00629)].\n\nAutomatic approaches can be divided into methods based on remote sensing time series and methods based on near- or mid-range imagery. Time-series methods use spectral-index dynamics, threshold rules, seasonal-curve approximations, and machine learning models such as Support Vector Machines (SVMs) and k-Nearest Neighbors (kNNs) [[4](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B4-drones-10-00629),[28](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B28-drones-10-00629)], as well as deep learning and transformer-based architectures [[29](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B29-drones-10-00629),[30](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B30-drones-10-00629)].\n\nPhenological normalization of the time axis can improve model portability [[31](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B31-drones-10-00629)]. However, satellite-based approaches remain limited by spatial resolution, cloudiness, and mixed pixels, especially in heterogeneous fields and small plots.\n\nImage-based approaches determine phenology from plant morphology and canopy structure using ground cameras, PhenoCam systems, mobile imagery, or UAV data. PhenoCam studies combine CNN classification with temporal post-processing such as Hidden Markov Models to obtain robust daily field-state sequences [[32](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B32-drones-10-00629),[33](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B33-drones-10-00629)]. UAV imagery provides lower temporal continuity but much higher spatial resolution, enabling plant- or row-level analysis and detection of within-field heterogeneity [[16](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B16-drones-10-00629),[34](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B34-drones-10-00629)]. Image-based phenological tasks include scene classification [[35](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B35-drones-10-00629)], organ detection [[24](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B24-drones-10-00629)], and segmentation-based analysis [[20](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B20-drones-10-00629)].\n\n### 2.2. Machine Learning Methods for Phenological Classification\n\nEarly automated phenological classification relied on manually designed spectral, textural, and canopy features followed by classical machine learning methods, including SVM and kNN [[28](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B28-drones-10-00629)]. Such methods can perform well under stable conditions, for example, in sugarcane stage identification from Sentinel-2 vegetation-index time series [[36](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B36-drones-10-00629)], but their transferability between regions and seasons is limited by differences in cultivars, management, weather, and phenological timing.\n\nRecent studies increasingly use deep learning, attention mechanisms, and temporal models. For UAV orthomosaics, MaizeHT combines CNNs and self-attention and reaches 97.7–98.7% accuracy in maize growth-stage recognition [[35](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B35-drones-10-00629)]. ResNet-50 has been applied to rice growth-stage classification from ground images [[37](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B37-drones-10-00629)], while CNN-LSTM and CNN-GRU models are used when image sequences are available [[38](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B38-drones-10-00629)]. For PhenoCam imagery, DL classification combined with HMM post-processing improves temporal consistency under management events such as plowing, harvesting, and crop rotation [[32](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B32-drones-10-00629)].\n\n### 2.3. Phenological Classification Using UAV Data and Deep Learning\n\nFor the present study, the most relevant works are those that classify phenological stages directly from UAV RGB or multispectral imagery using deep learning. Reviews confirm that UAVs are valuable for such tasks because they combine high spatial resolution, flexible survey timing, and operation below cloud cover [[17](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B17-drones-10-00629),[18](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B18-drones-10-00629)]. However, adjacent phenological stages often differ only in subtle morphological details, so classification errors commonly occur between neighboring classes [[39](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B39-drones-10-00629),[40](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B40-drones-10-00629)].\n\nA further limitation concerns data and labeling. Many studies rely on local datasets collected in one region with a fixed sensor and survey protocol, which may overestimate performance because of hidden correlations in background, illumination, and management conditions. Consequently, practical deployment requires attention to domain shifts and stronger evaluation protocols such as leave-one-field or leave-one-season-out validation [[41](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B41-drones-10-00629)].\n\nIn Central Asia, including Kazakhstan, UAV-based phenological classification remains underrepresented despite the region’s agro-climatic diversity and the importance of grain and oilseed production. A relevant dataset for wheat, barley, and soybean in East Kazakhstan has been published using multispectral UAV imagery [[42](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B42-drones-10-00629)]. Related studies from Kazakhstan also show the potential of UAVs and machine learning methods for soybean plant and weed detection [[24](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B24-drones-10-00629)] and agricultural salinity mapping [[13](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B13-drones-10-00629)], but available datasets remain limited in crops, years, and survey conditions. Standardized BBCH labeling is therefore important for comparability and for representing mixed field states [[25](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B25-drones-10-00629)].\n\nOverall, recent research highlights several persistent limitations:\n\n*   Weak separability of adjacent stages and continuous transitions between phases [[36](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B36-drones-10-00629),[38](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B38-drones-10-00629),[40](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B40-drones-10-00629)];\n\n*   Heterogeneous crop states within a field, which complicate the assumption of one stage per image [[28](https:\u002F\u002Fwww.mdpi.com\u002F2504-446X\u002F10\u002F8\u002F629#B28-drones-10-00629)];\n\n*   Limited availability of labeled BBCH datasets covering diffe","MDPI Drones 2026-08-17","2026-08-17T00:00:00Z",{"impact":19,"substance":18,"depth":16,"authority":19,"freshness":12,"relevant":21,"comment":450},"论文提出基于无人机RGB影像和迁移学习的作物物候期分类方法，数据规模大、精度高，对精准农业有参考价值，但属细分领域进展。",[452],{"name":447,"url":444},[106,454,455],"UAV ResNet18","物候期分类",{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},[458],{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},{"status":45,"locked":13,"version":46,"method":47,"evidence":460},{"candidates":461,"agricultural_evidence":52},[462,466,472,476],{"code":39,"score":51,"supported":52,"title":463,"abstract":464,"tags":465},[117],[119],[],{"code":32,"score":59,"supported":52,"title":467,"abstract":468,"tags":471},[],[56,469,470],"油菜","向日葵",[],{"code":128,"score":201,"supported":52,"title":473,"abstract":474,"tags":475},[],[106],[106],{"code":477,"score":207,"supported":52,"title":478,"abstract":479,"tags":481},"A2",[],[480],"小麦",[],[483,484],"物候期分类 精准农业 UAV ResNet18","物候期分类 精准农业","物候期分类精准农业UAVResNet18-676",5,"2026-08-17T23:54:56.040561Z"]