[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-2901":3,"related-2901":70},{"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":22,"tags":24,"primary_topic":8,"secondary_topics":30,"classification":31,"search_phrases":55,"slug":58,"view_count":39,"doi":8,"paper":59,"created_at":69},2901,"AgriScope:面向农业图像的像素级多模态理解统一框架,arXiv 2609.20325(预印本)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.20325","Boudiaf、Alanssari、Hussain、Javed提出AgriScope,一个统一的像素级多模态农业图像理解框架,联合支持图像级、区域级、像素级理解,可实现接地描述生成、指代表达分割、多轮多模态交互等任务。集成生物专用语义表征、密集空间表征与像素解码;引入大规模像素级农业多模态指令调优数据集AgriGround,包含50万+图像和1100万+指令跟随样本,涵盖植物病害分析、作物与杂草识别、昆虫识别、细粒度植物理解。实验表明AgriScope在多项农业视觉语言任务上有效。",null,"arXiv (preprint)","2026-09-17T00:00:00Z","论文",10,false,75,{"impact":16,"substance":17,"depth":16,"authority":18,"freshness":19,"relevant":20,"comment":21},18,22,8,9,1,"提出统一像素级农业多模态理解框架并开源50万图像、1100万指令样本的大规模数据集，方法新颖、数据规模突出，但为arXiv预印本、未经同行评审，权威性有限，值得作为前沿技术动态精选。",[23],{"name":9,"url":6},[25,26,27,28,29],"智慧农业","农业人工智能","农业遥感","植物病害识别","多模态大模型",[],{"status":32,"locked":13,"version":33,"method":34,"evidence":35},"unclassified","2026-10-22.2","rules",{"candidates":36,"agricultural_evidence":53,"reason":54},[37,43,48],{"code":38,"score":39,"supported":13,"title":40,"abstract":41,"tags":42},"I2",3,[],[],[26,29,28],{"code":44,"score":20,"supported":13,"title":45,"abstract":46,"tags":47},"I1",[],[],[25],{"code":49,"score":20,"supported":13,"title":50,"abstract":51,"tags":52},"I3",[],[],[27],true,"缺少足够的主题证据",[56,57],"AgriScope 农业图像 多模态","AgriGround 像素级 农业数据集","AgriScope农业图像多模态-2901",{"doi":8,"openalex_id":8,"authors":60,"venue":8,"cited_by_count":61,"oa_url":8,"card":62,"direction":66,"ingested_from":68},[],0,{"tldr":63,"method":64,"finding":65,"direction":66,"opportunity":67},"提出AgriScope统一框架，实现农业图像像素级多模态理解与多任务交互。","构建AgriGround数据集（50万+图像、1100万+指令样本），融合语义与","AgriScope在接地描述、指代分割、多轮交互等农业视觉语言任务上有效。","农业人工智能与决策模型","可探索像素级多模态模型在田间实时病害诊断与精准施药决策中的落地与轻量化。","agent","2026-09-19T00:06:08.678379Z",{"total":71,"page":20,"page_size":71,"items":72},6,[73,155,223,324,386,445],{"id":74,"title":75,"url":76,"summary":77,"summary_zh":78,"content":8,"source_name":79,"source_url":76,"published_at":80,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":81,"score_detail":82,"sources":85,"tags":87,"primary_topic":8,"secondary_topics":89,"classification":90,"search_phrases":126,"slug":129,"view_count":130,"doi":131,"paper":132,"created_at":154},3159,"A multi-agent multi-modal LLM with Monte Carlo tree search for interpretable plant disease classification","https:\u002F\u002Fdoi.org\u002F10.3389\u002Ffpls.2026.1937679","In plant disease classification tasks, existing task specific models for plant disease classification typically achieve high accuracy by employing tailored network architectures. However, these models are unable to provide reasoning chains, which reduces interpretability and undermines trustworthiness in practical applications. In addition, although multimodal large language models (MLLMs) have made significant progress in smart agriculture, their disease classification accuracy lags behind that of task-specific models due to insufficient domain knowledge. While fine-tuning can enhance domain-specific knowledge, relying solely on unimodal data or simple multimodal information remains insufficient for adequately improving classification accuracy. Specifically, the simple multimodal information in this context refers to a single global description of the entire image produced by an MLLM. To address these issues, this paper proposes a multi-agent collaborative classification framework that integrates Monte Carlo Tree Search (MCTS) with multimodal enhancement. With an MLLM as the core, the framework first introduces an MCTSguided multi-agent collaboration mechanism that directs the agents to progressively generate region-specific descriptive textual information, which together with the image forms complex multimodal information and thereby boosts classification performance. Meanwhile, the search trajectories generated by MCTS are leveraged to construct reasoning processes for enhanced interpretability. Finally, a self-training mechanism is adopted to iteratively optimize the model, further improving overall performance. Experimental results show that the framework improves the baseline MLLM’s accuracy while providing interpretable reasoning chains, achieving gains from 0.6701 to 0.9965 on 3-class potato, 0.5340 to 0.9651 on 5-class potato, 0.5387 to 0.9467 on tomato, and 0.4705 to 0.9948 on grape.","在植物病害分类任务中，现有的植物病害分类专用模型通常通过采用定制化的网络架构来实现高准确率。然而，这些模型无法提供推理链，这降低了可解释性，并在实际应用中削弱了可信度。此外，尽管多模态大语言模型（MLLMs）在智慧农业领域取得了显著进展，但由于领域知识不足，其病害分类准确率仍落后于专用模型。虽然微调可以增强领域特定知识，但仅依赖单模态数据或简单的多模态信息仍不足以充分提升分类准确率。具体而言，此处的简单多模态信息是指MLLM对整个图像生成的单一全局描述。为解决上述问题，本文提出了一种融合蒙特卡洛树搜索（MCTS）与多模态增强的多智能体协同分类框架。该框架以MLLM为核心，首先引入MCTS引导的多智能体协同机制，引导各智能体逐步生成区域特定的描述性文本信息，这些信息与图像共同构成复杂的多模态信息，从而提升分类性能。同时，利用MCTS生成的搜索轨迹构建推理过程，以增强可解释性。最后，采用自训练机制迭代优化模型，进一步提升整体性能。实验结果表明，该框架在提升基线MLLM准确率的同时提供了可解释的推理链，在三分类马铃薯上从0.6701提升至0.9965，在五分类马铃薯上从0.5340提升至0.9651，在番茄上从0.5387提升至0.9467，在葡萄上从0.4705提升至0.9948。","Frontiers in Plant Science","2026-09-21T00:00:00Z",80,{"impact":16,"substance":17,"depth":16,"authority":83,"freshness":18,"relevant":20,"comment":84},14,"提出多智能体+MCTS+多模态增强的可解释植物病害分类框架，准确率大幅提升，方法新颖且结论可靠，对智慧农业AI应用有参考价值。",[86],{"name":79,"url":76},[25,26,88,28,29],"可解释AI",[],{"status":91,"locked":13,"version":33,"method":34,"evidence":92},"needs_review",{"candidates":93,"agricultural_evidence":53,"reason":125},[94,100,107,114,121],{"code":44,"score":95,"supported":53,"title":96,"abstract":97,"tags":99},7,[],[25,98],"smart agriculture",[25],{"code":101,"score":71,"supported":53,"title":102,"abstract":103,"tags":106},"A5",[],[104,105],"马铃薯","potato",[],{"code":108,"score":71,"supported":53,"title":109,"abstract":110,"tags":113},"B1",[],[111,112],"番茄","tomato",[],{"code":115,"score":71,"supported":53,"title":116,"abstract":117,"tags":120},"B2",[],[118,119],"葡萄","grape",[],{"code":38,"score":39,"supported":13,"title":122,"abstract":123,"tags":124},[],[],[26,88,29,28],"多个产业证据相当，需要确认主归属",[127,128],"多智能体 蒙特卡洛树搜索 植物病害分类","多模态大模型 马铃薯 番茄 葡萄 病害识别","多智能体蒙特卡洛树搜索植物病害分类-3159",2,"10.3389\u002Ffpls.2026.1937679",{"doi":131,"openalex_id":133,"authors":134,"venue":79,"cited_by_count":61,"oa_url":76,"card":147,"direction":152,"ingested_from":153},"W7213924931",[135,138,140,142,145],{"name":136,"orcid":137},"Ridong Wu","https:\u002F\u002Forcid.org\u002F0000-0003-0729-3099",{"name":139,"orcid":8},"Hua Zou",{"name":141,"orcid":8},"Zhijie Li",{"name":143,"orcid":144},"Qian Zhou","https:\u002F\u002Forcid.org\u002F0000-0001-7964-8130",{"name":146,"orcid":8},"一子 江﨑",{"tldr":148,"method":149,"finding":150,"direction":66,"opportunity":151},"提出多智能体多模态LLM结合蒙特卡洛树搜索，实现可解释的植物病害分类。","MCTS引导多智能体生成区域描述文本，构建复杂多模态信息并自训练优化。","分类准确率大幅提升，如马铃薯三分类从0.6701升至0.9965，并提供推理链。","可探索将MCTS多智能体框架迁移至其他作物病害或田间复杂场景，提升可解释性与泛化性。","智慧农业 \u002F 农业物联网","openalex","2026-09-22T23:30:11.045394Z",{"id":156,"title":157,"url":158,"summary":159,"summary_zh":160,"content":8,"source_name":161,"source_url":158,"published_at":162,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":163,"score_detail":164,"sources":168,"tags":170,"primary_topic":173,"secondary_topics":179,"classification":185,"search_phrases":203,"slug":206,"view_count":61,"doi":207,"paper":208,"created_at":222},5592,"Evaluation of Concept-Rank Collapse Audits for Computer Vision Agriculture Disease Classifiers Under Low-Label Regimes: An Interpretability Capacity Bound for Smallholder Fungicide Selectivity","https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.23545193","Concept-based interpretability methods for plant-disease classifiers are typically evaluated by whether their concept activations correlate with human-annotated labels, but this criterion is silent on a prior question: whether the classifier's internal representation retains enough concept-rank structure for any audit to be informative. the present analysis formalises that prior question as an interpretability capacity bound. A classifier is modelled as a concept-rank operator whose singular values govern how much concept-level signal survives the projection from pixel space to class logits, and a collapse audit is defined as the ratio of retained concept-rank energy to total concept energy. Under a low-label regime, where concept annotations are scarce and the audit must be estimated from few labelled examples, the bound predicts that audit reliability degrades as a function of the concept-rank spectrum and the label budget. The analysis is a theoretical and methodological contribution: it derives the bound, states the conditions under which a collapse audit is identifiable, and characterises the operating characteristics of the audit as a decision procedure under a declared variance structure (between-cluster SD = 6.0, residual SD = 4.5, intraclass correlation = 0.64). Power was estimated across 2000 design cells for cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points; 52 clusters were required for 5 percentage points at 80% power. The bound reframes concept audits as capacity tests rather than fidelity tests, with consequences for how smallholder fungicide-selectivity decisions should be supported by interpretable disease classifiers.","面向植物病害分类器的基于概念的可解释性方法，通常以其概念激活是否与人工标注标签相关作为评价标准，但该标准对一个更先前的问题保持沉默：分类器的内部表征是否保留了足够的概念秩结构，从而使任何审计具有信息量。本分析将该先前问题形式化为一个可解释性容量上界。分类器被建模为概念秩算子，其奇异值支配着从像素空间到类别logits的投影中概念级信号得以保留的程度，并将坍缩审计定义为保留的概念秩能量与总概念能量之比。在低标签情形下，概念标注稀缺且审计必须由少量标注样本估计，该上界预测审计可靠性会随概念秩谱和标签预算的函数而退化。本分析是一项理论和方法论贡献：推导出该上界，陈述坍缩审计可识别的条件，并在声明的方差结构（簇间SD = 6.0，残差SD = 4.5，组内相关系数 = 0.64）下刻画审计作为决策程序的操作特征。在2000个设计单元中，针对簇数为8、16、24和32以及效应量为3、5和8个百分点的情形估计了功效；在80%功效下达到5个百分点需要52个簇。该上界将概念审计重新框定为容量检验而非保真度检验，并对可解释病害分类器应如何支持小农杀菌剂选择性决策具有启示意义。","Zenodo (CERN European Organization for Nuclear Research)","2026-10-21T00:00:00Z",69,{"impact":18,"substance":165,"depth":16,"authority":166,"freshness":12,"relevant":20,"comment":167},21,12,"理论方法论文，提出可解释性容量上界并给出审计功效分析，对智慧农业病害识别可信度有参考价值，但属小众方法学研究、缺乏田间实证，暂不宜进入每日精选。",[169],{"name":161,"url":158},[25,26,171,28,172],"小农户","可解释性",{"code":174,"slug":175,"name":176,"path":177,"parent_code":178},"G2","farm-services","社会化服务与规模经营","\u002Ftopic\u002Frural-services\u002Ffarm-services","G",[180],{"code":38,"slug":181,"name":182,"path":183,"parent_code":184},"agricultural-ai","农业AI与大模型","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-ai","I",{"status":186,"locked":13,"version":33,"method":34,"evidence":187},"classified",{"candidates":188,"agricultural_evidence":53},[189,194,199],{"code":174,"score":166,"supported":53,"title":190,"abstract":192,"tags":193},[191],"smallholder",[191],[171],{"code":38,"score":12,"supported":53,"title":195,"abstract":197,"tags":198},[196],"computer vision",[],[26,28],{"code":44,"score":20,"supported":13,"title":200,"abstract":201,"tags":202},[],[],[25],[204,205],"农业人工智能 植物病害识别 可解释性 智慧农业","农业人工智能 植物病害识别","农业人工智能植物病害识别可解释性智慧农业-5592","10.5281\u002Fzenodo.23545193",{"doi":207,"openalex_id":209,"authors":210,"venue":161,"cited_by_count":61,"oa_url":158,"card":217,"direction":66,"ingested_from":153},"W7223137323",[211,213,215],{"name":212,"orcid":8},"Akshay Ramgoolam",{"name":214,"orcid":8},"Priya Bundhoo Appadoo",{"name":216,"orcid":8},"Jean-Noël LEspérance",{"tldr":218,"method":219,"finding":220,"direction":66,"opportunity":221},"提出可解释性容量上界，评估低标签下概念秩崩溃审计的可靠性。","将分类器建模为概念秩算子，推导容量上界并做2000格功效模拟。","审计可靠性随概念秩谱与标签预算下降；5个百分点效应需52个聚类达80%功效。","可将容量上界用于小农户杀菌剂选择性决策的可解释模型审计与标签预算设计。","2026-10-22T11:21:10.425290Z",{"id":224,"title":225,"url":226,"summary":227,"summary_zh":228,"content":8,"source_name":229,"source_url":226,"published_at":162,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":81,"score_detail":230,"sources":233,"tags":235,"primary_topic":238,"secondary_topics":244,"classification":255,"search_phrases":291,"slug":294,"view_count":61,"doi":295,"paper":296,"created_at":323},5573,"Improving alfalfa nitrogen estimation with canopy structural priors from DSSAT–PROSAIL and UAV hyperspectral data","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.eja.2026.128375","Accurate estimation of nitrogen (N) concentration in alfalfa ( Medicago sativa L .) supports forage quality assessment and harvest scheduling. Empirical remote sensing models often overfit because of confounding effects of three-dimensional canopy structure. Incorporating structural parameters can stabilize biochemical inversions, but concurrent in situ morphological measurements are difficult to acquire at scale. We developed the Morphology-Aware Latent Embedding Network (MALENet), a hybrid framework integrating physiological and morphological mechanisms with hyperspectral data. The Decision Support System for Agrotechnology Transfer (DSSAT), coupled with the PROSAIL radiative transfer model, generated a structurally rich synthetic hyperspectral prior. A convolutional neural network (CNN) with a Squeeze-and-Excitation (SE) attention mechanism, was pre-trained on these data to distill ecophysiological knowledge into a dense latent representation. This mechanistically derived embedding was then fused with empirical hyperspectral measurements collected under diverse field conditions. Evaluation using multi-year field quadrats in the Hexi Corridor showed improved predictive performance ( R² = 0.71, RMSE = 0.30%) relative to conventional machine learning and deep learning models (LASSO-ANN: R² = 0.61; CNN: R² = 0.66). Feature attribution indicated that MALENet used near-infrared multiple scattering (58.3% attribution) as a structural proxy to calibrate nitrogen concentration retrieval against the confounding effects of canopy architecture. Despite constraints imposed by sample size and PROSAIL's 1D canopy assumption, these findings demonstrate the potential of physically guided deep learning to use structural information for agricultural trait retrieval in precision forage management.","准确估算紫花苜蓿（Medicago sativa L.）的氮浓度有助于饲草品质评估和收获期安排。经验性遥感模型常因三维冠层结构的混杂效应而出现过拟合。引入结构参数可以稳定生化反演，但同步开展原位形态学测量难以大规模获取。我们开发了形态感知潜在嵌入网络（Morphology-Aware Latent Embedding Network，MALENet），这是一个将生理与形态机制同高光谱数据相结合的混合框架。农业技术转移决策支持系统（Decision Support System for Agrotechnology Transfer，DSSAT）与PROSAIL辐射传输模型耦合，生成了结构信息丰富的合成高光谱先验数据。在此基础上，预训练了一个带有挤压—激励（Squeeze-and-Excitation，SE）注意力机制的卷积神经网络（convolutional neural network，CNN），将生态生理学知识提炼为稠密潜在表征。随后，将这一机制性导出的嵌入与在不同田间条件下采集的经验性高光谱测量数据相融合。利用河西走廊多年田间样方进行的评估显示，相较于传统机器学习和深度学习模型（LASSO-ANN：R² = 0.61；CNN：R² = 0.66），预测性能有所提升（R² = 0.71，RMSE = 0.30%）。特征归因表明，MALENet利用近红外多次散射（归因占比58.3%）作为结构代理，以校准氮浓度反演，抵御冠层结构的混杂效应。尽管受样本量和PROSAIL一维冠层假设的限制，这些发现表明，物理引导的深度学习有潜力利用结构信息，在精准饲草管理中进行农业性状反演。","European Journal of Agronomy",{"impact":231,"substance":17,"depth":16,"authority":83,"freshness":12,"relevant":20,"comment":232},16,"该研究将作物模型与辐射传输模型耦合生成结构先验，并结合无人机高光谱数据反演苜蓿氮素，方法新颖、田间验证扎实，对精准饲草管理有参考价值，值得进入每日精选。",[234],{"name":229,"url":226},[25,26,27,236,237],"氮素监测","紫花苜蓿",{"code":239,"slug":240,"name":241,"path":242,"parent_code":243},"A7","forage-cropping","饲草与耕作制度","\u002Ftopic\u002Ffield-crops\u002Fforage-cropping","A",[245,249],{"code":49,"slug":246,"name":247,"path":248,"parent_code":184},"remote-sensing","遥感与农情监测","\u002Ftopic\u002Fdigital-agriculture\u002Fremote-sensing",{"code":250,"slug":251,"name":252,"path":253,"parent_code":254},"C4","feed","饲料与饲草","\u002Ftopic\u002Flivestock\u002Ffeed","C",{"status":186,"locked":13,"version":33,"method":34,"evidence":256},{"candidates":257,"agricultural_evidence":53},[258,265,272,278,287],{"code":49,"score":16,"supported":53,"title":259,"abstract":261,"tags":264},[260],"hyperspectral",[262,263,260],"遥感","remote sensing",[27],{"code":239,"score":266,"supported":53,"title":267,"abstract":269,"tags":271},15,[268],"alfalfa",[270,268],"苜蓿",[237],{"code":250,"score":12,"supported":53,"title":273,"abstract":274,"tags":277},[],[275,270,276],"饲草","forage",[237],{"code":38,"score":12,"supported":53,"title":279,"abstract":280,"tags":286},[],[281,282,283,284,285],"深度学习","机器学习","machine learning","deep learning","decision support",[26],{"code":44,"score":20,"supported":13,"title":288,"abstract":289,"tags":290},[],[],[25],[292,293],"DSSAT PROSAIL 苜蓿 氮素","无人机 高光谱 苜蓿","DSSATPROSAIL苜蓿氮素-5573","10.1016\u002Fj.eja.2026.128375",{"doi":295,"openalex_id":297,"authors":298,"venue":229,"cited_by_count":61,"oa_url":226,"card":317,"direction":321,"ingested_from":153},"W7223154576",[299,301,303,305,307,310,312,315],{"name":300,"orcid":8},"Shuai Fu",{"name":302,"orcid":8},"Yunhao Li",{"name":304,"orcid":8},"Chunli Miao",{"name":306,"orcid":8},"Jie Liu",{"name":308,"orcid":309},"Tiangang Liang","https:\u002F\u002Forcid.org\u002F0000-0001-5599-6587",{"name":311,"orcid":8},"Yongjia Hu",{"name":313,"orcid":314},"Qisheng Feng","https:\u002F\u002Forcid.org\u002F0009-0000-1391-9591",{"name":316,"orcid":8},"吴彩霞",{"tldr":318,"method":319,"finding":320,"direction":321,"opportunity":322},"构建MALENet融合DSSAT-PROSAIL结构先验与无人机高光谱，提升苜蓿氮浓度估算精度。","DSSAT耦合PROSAIL生成合成高光谱先验，预训练CNN-SE网络，再融合田","MALENet精度优于传统机器学习和深度学习模型，R²达0.71，近红外多次散射作为结构代理贡献58","农业遥感与作物表型","可探索将作物模型结构先验迁移至其他牧草或作物生化参数反演，并缓解PROSAIL一维冠层假设局限。","2026-10-22T11:20:39.622844Z",{"id":325,"title":326,"url":327,"summary":328,"summary_zh":329,"content":8,"source_name":330,"source_url":327,"published_at":331,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":332,"score_detail":333,"sources":335,"tags":337,"primary_topic":340,"secondary_topics":341,"classification":342,"search_phrases":353,"slug":356,"view_count":61,"doi":357,"paper":358,"created_at":385},5562,"pxGPT: A schema-constrained multimodal LLM framework for plant phenotyping","https:\u002F\u002Fdoi.org\u002F10.1093\u002Fbioinformatics\u002Fbtag781","Abstract Summary Deep learning can scale image-based plant phenotyping but typically targets predefined traits, whereas manual phenotypic description is more flexible but difficult to scale. We present Phenotype eXplorer Generative Plant Traits (pxGPT), a Python framework that converts multi-view plant images into structured phenotype records. The workflow combines free-text description, dataset-specific schema generation with human review, and schema-constrained information extraction. We evaluated the extraction stage on three Brassica datasets using 78,973 annotations from ten annotators who had no access to pxGPT predictions. For 38-45 qualitative traits per dataset, pxGPT-human agreement ranged from 0.572 to 0.712, compared with 0.620 to 0.739 agreement among human annotators. All 32,736 required trait fields were present; 93 values were empty, and no qualitative descriptor violations were recorded. For selected morphological structures, pooled false-positive rates were 0.5-8.0%, while sensitivity varied across structure types in one mature-plant dataset (37.5-100%). The curated schemas were applied unchanged with two commercial models and one locally hosted open-weight model, with differences in mean accuracy of 0.03 or less. In the evaluated Brassica datasets, these results support schema-guided extraction of visible plant traits without separate model training for each trait. Human-guided schema development offers a path to extend pxGPT to other crops and imaging conditions, with accuracy requiring validation in each new setting. Availability and Implementation The software is available at https:\u002F\u002Fgithub.com\u002FYaoChengLab\u002Fpxgpt and an archival version of the source code is available on Zenodo (https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.22931745). Curated schemas used in this study, annotation guides, phenotype records and human annotations are available at https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.22907554. Supplementary information Supplementary data are available at Bioinformatics online.","摘要 深度学习能够扩展基于图像的植物表型分析，但通常针对预定义性状，而人工表型描述更为灵活却难以规模化。我们提出Phenotype eXplorer Generative Plant Traits（pxGPT），一个将多视角植物图像转换为结构化表型记录的Python框架。该工作流程结合了自由文本描述、数据集特定模式生成与人工审核，以及模式约束的信息提取。我们在三个芸薹属（Brassica）数据集上评估了提取阶段，使用了来自十名标注者的78,973条标注，这些标注者无法获取pxGPT的预测结果。对于每个数据集38-45个定性性状，pxGPT与人工标注者之间的一致性范围为0.572至0.712，而人工标注者之间的一致性为0.620至0.739。所有32,736个所需性状字段均已存在；93个值为空，未记录到定性描述符违规。对于选定的形态结构，合并假阳性率为0.5-8.0%，而在一个成熟植株数据集中，灵敏度因结构类型而异（37.5-100%）。经整理的模式在未作修改的情况下应用于两个商业模型和一个本地部署的开放权重模型，平均准确率差异不超过0.03。在所评估的芸薹属数据集中，这些结果支持在无需为每个性状单独训练模型的情况下，对可见植物性状进行模式引导的提取。人工引导的模式开发为将pxGPT扩展至其他作物和成像条件提供了途径，其准确性需在每个新设置中加以验证。可用性与实现 该软件可在https:\u002F\u002Fgithub.com\u002FYaoChengLab\u002Fpxgpt获取，源代码的存档版本可在Zenodo（https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.22931745）获取。本研究中使用的经整理模式、标注指南、表型记录和人工标注可在https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.22907554获取。补充信息 补充数据可在Bioinformatics在线获取。","Bioinformatics","2026-10-20T00:00:00Z",81,{"impact":16,"substance":17,"depth":16,"authority":83,"freshness":19,"relevant":20,"comment":334},"提出基于模式约束的多模态大模型植物表型提取框架，在近8万条标注上验证，方法新颖、数据规模可观，对智慧育种与表型自动化有参考价值。",[336],{"name":330,"url":327},[25,26,338,29,339],"植物表型","芸薹属",{"code":38,"slug":181,"name":182,"path":183,"parent_code":184},[],{"status":186,"locked":13,"version":33,"method":34,"evidence":343},{"candidates":344,"agricultural_evidence":53},[345,349],{"code":38,"score":18,"supported":53,"title":346,"abstract":347,"tags":348},[],[281,284],[26,29],{"code":44,"score":20,"supported":13,"title":350,"abstract":351,"tags":352},[],[],[25],[354,355],"pxGPT 植物表型","Brassica 表型 多模态大模型","pxGPT植物表型-5562","10.1093\u002Fbioinformatics\u002Fbtag781",{"doi":357,"openalex_id":359,"authors":360,"venue":330,"cited_by_count":61,"oa_url":379,"card":380,"direction":152,"ingested_from":153},"W7223127960",[361,364,366,369,372,375,377],{"name":362,"orcid":363},"Po‐Xing Zheng","https:\u002F\u002Forcid.org\u002F0000-0001-5827-5865",{"name":365,"orcid":8},"Jheng-Yan Ou",{"name":367,"orcid":368},"Yu-Lin Wu","https:\u002F\u002Forcid.org\u002F0009-0003-8596-9047",{"name":370,"orcid":371},"Ruei-Shiuan Wang","https:\u002F\u002Forcid.org\u002F0009-0007-5159-1129",{"name":373,"orcid":374},"WU Ting-li","https:\u002F\u002Forcid.org\u002F0000-0001-7996-5515",{"name":376,"orcid":8},"Te-Chang Hsu",{"name":378,"orcid":8},"Yao-Cheng Lin","https:\u002F\u002Facademic.oup.com\u002Fbioinformatics\u002Fadvance-article-pdf\u002Fdoi\u002F10.1093\u002Fbioinformatics\u002Fbtag781\u002F72201337\u002Fbtag781.pdf",{"tldr":381,"method":382,"finding":383,"direction":321,"opportunity":384},"提出pxGPT框架，用多模态大模型将多视角植物图像转为结构化表型记录。","结合自由文本描述、人工审核的模式生成与模式约束信息抽取，在三个芸薹属数据集上评估","pxGPT与人工标注一致性0.572-0.712，接近人工间一致性，且无需逐性状训练模型。","可探索将模式引导抽取迁移至更多作物与复杂成像条件，并研究自动化模式生成以降低人工审核成本。","2026-10-22T11:20:22.173561Z",{"id":387,"title":388,"url":389,"summary":390,"summary_zh":391,"content":8,"source_name":161,"source_url":389,"published_at":392,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":393,"score_detail":394,"sources":396,"tags":398,"primary_topic":401,"secondary_topics":402,"classification":403,"search_phrases":421,"slug":424,"view_count":20,"doi":425,"paper":426,"created_at":444},5194,"A Comprehensive Review on Multi-Modal Large Language Models and Agentic Architectures for Precision Agricultural Decision Support","https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.23411181","This article reviews recent developments in multi-modal large language models and agentic architectures for agricultural decision support. Forty-five representative studies published between 2023 and 2026 were selected across perception, language, orchestration, soil intelligence and governance, organised under a four-layer taxonomy, and classified against nine functional capabilities using a stated, reproducible procedure. The comparative analysis exposes a pronounced imbalance. Crop and disease perception is covered by 36 per cent of the selected studies, while dose-bearing prescription appears in 7 per cent, profit-aware recommendation in 4 per cent, and persistent per-field memory in a single study. Advisory tools in this literature therefore remain unable to exchange state: image-based scanners are rarely conditioned on measured soil chemistry, and crop-recommendation models optimise agronomic suitability without reference to farm economics. On this basis the article synthesises a conceptual reference architecture from the reviewed literature. It is a literature-derived synthesis rather than an implemented system, and calibrated abstention, evaluation protocols for longitudinal memory, and vernacular voice coverage emerge as the most urgent gaps.","本文综述了面向农业决策支持的多模态大语言模型与智能体架构的最新进展。按照可复现的既定流程，选取2023年至2026年间发表的45项代表性研究，涵盖感知、语言、编排、土壤智能与治理等方向，将其纳入一个四层分类体系，并依据九项功能能力进行分类。比较分析揭示出明显的不均衡：作物与病害感知在所选研究中占36%，而含剂量处方的研究仅占7%，利润感知推荐占4%，持久性田块记忆仅见于一项研究。因此，该领域的咨询工具仍无法交换状态：基于图像的扫描器很少以实测土壤化学状况为条件，作物推荐模型在优化农艺适宜性时也未考虑农场经济性。在此基础上，本文综合所综述文献提出一个概念性参考架构。该架构是文献衍生的综合框架，而非已实现的系统；校准弃权、纵向记忆的评估协议以及方言语音覆盖被列为最紧迫的空白。","2026-10-16T00:00:00Z",63,{"impact":166,"substance":16,"depth":231,"authority":18,"freshness":19,"relevant":20,"comment":395},"系统性综述揭示处方决策与农场经济维度严重缺位，对农业AI研发方向有参考价值，但属文献综述而非实证突破，影响范围有限。",[397],{"name":161,"url":389},[25,26,399,29,400],"智能体","农业决策支持",{"code":38,"slug":181,"name":182,"path":183,"parent_code":184},[],{"status":186,"locked":13,"version":33,"method":34,"evidence":404},{"candidates":405,"agricultural_evidence":53},[406,410,417],{"code":38,"score":83,"supported":53,"title":407,"abstract":408,"tags":409},[285],[285],[26,29,399],{"code":411,"score":71,"supported":53,"title":412,"abstract":413,"tags":416},"H2",[],[414,415],"土壤","soil",[],{"code":44,"score":20,"supported":13,"title":418,"abstract":419,"tags":420},[],[],[25],[422,423],"多模态大模型 农业决策支持","农业人工智能 农业决策支持 多模态大模型 智慧农业","多模态大模型农业决策支持-5194","10.5281\u002Fzenodo.23411181",{"doi":425,"openalex_id":427,"authors":428,"venue":161,"cited_by_count":61,"oa_url":389,"card":439,"direction":66,"ingested_from":153},"W7222605322",[429,431,433,435,437],{"name":430,"orcid":8},"Tanmay Bhamre",{"name":432,"orcid":8},"Yash Barde",{"name":434,"orcid":8},"Devendra Yeole",{"name":436,"orcid":8},"Gajanan Chaudhary",{"name":438,"orcid":8},"Amarsinh Verpe",{"tldr":440,"method":441,"finding":442,"direction":66,"opportunity":443},"综述多模态大模型与智能体架构在精准农业决策支持中的应用与不足。","筛选2023-2026年45篇代表性研究，按四层分类与九项功能能力归类比较。","感知研究占36%，处方仅7%、利润推荐4%、田块记忆仅1篇，系统难以交换状态。","可研究跨模态状态交换、纵向记忆评测协议与方言语音覆盖，填补处方与经济决策空白。","2026-10-17T23:30:55.282694Z",{"id":446,"title":447,"url":448,"summary":449,"summary_zh":450,"content":8,"source_name":451,"source_url":448,"published_at":392,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":452,"score_detail":453,"sources":457,"tags":459,"primary_topic":462,"secondary_topics":463,"classification":464,"search_phrases":483,"slug":486,"view_count":61,"doi":487,"paper":488,"created_at":502},5192,"A Comparative Study of Client-Level Stabilization Techniques in Federated Learning for Robust Plant Disease Recognition","https:\u002F\u002Fdoi.org\u002F10.3390\u002Fs26206566","The successful integration of artificial intelligence (AI) algorithms in precision agriculture provided farmers with the means to identify potential issues, such as early disease detection or pest infestations, by making crop-related predictions. The new technological developments in this field improve the functioning of traditional agriculture by reducing the dependence on specialized human expertise and facilitating greater access to diagnostic procedures. However, most studies conducted in this area rely on centralized training of AI models, which leads to increased costs and raised privacy-related concerns. Federated learning (FL) offers a compelling framework for collaboratively training machine learning (ML) models on distributed datasets without centralizing the data owned by each participant. FL offers significant advantages in the preservation of data privacy, facilitates the collaborative exploitation of geographically distributed data sources, and reduces the risks associated with centralized collection. In this paper, we investigate the current limitations and performance boundaries of existing Federated and privacy-preserving learning in distributed sensing systems for plant disease detection, in scenarios that closely resemble real-world conditions. Our analysis specifically examined how FL aggregation algorithms, such as FedAvg, FedAvgM and FedProx, work in a custom multi-client experimental setup where data is partitioned so that each participant observes only those symptom patterns associated with specific environmental and climatic conditions. We propose further partial aggregation of the local model, an approach where we send only the model backbone to the central server, while the classification head of each participant remains strictly on the client side. The construction of environmentally heterogeneity is combined with a globally representative stabilizer node designed to preserve knowledge across disease classes that may be absent from certain clients.","人工智能（AI）算法在精准农业中的成功集成，为农民提供了通过作物相关预测来识别潜在问题（如早期病害检测或虫害侵染）的手段。该领域的新技术发展通过减少对专业人类知识的依赖并促进更广泛的诊断程序获取，改善了传统农业的运作方式。然而，该领域的大多数研究依赖于AI模型的集中式训练，这导致成本增加并引发隐私相关担忧。联邦学习（FL）提供了一个引人注目的框架，可在分布式数据集上协作训练机器学习（ML）模型，而无需集中化各参与者拥有的数据。联邦学习在保护数据隐私方面具有显著优势，促进了地理上分布的数据源的协作利用，并降低了与集中收集相关的风险。本文研究了在高度接近真实条件的场景中，现有联邦学习与隐私保护学习在用于植物病害检测的分布式感知系统中的当前局限性与性能边界。我们的分析特别考察了联邦学习聚合算法（如FedAvg、FedAvgM和FedProx）在自定义多客户端实验设置中的工作方式，其中数据被划分，使得每个参与者仅观察到与特定环境和气候条件相关的症状模式。我们提出对本地模型进行进一步的部分聚合，即仅将模型骨干发送至中央服务器，而每个参与者的分类头严格保留在客户端侧。环境异质性的构建与一个全局代表性的稳定器节点相结合，该节点旨在保留某些客户端可能缺失的跨病害类别的知识。","Sensors",78,{"impact":16,"substance":165,"depth":454,"authority":455,"freshness":19,"relevant":20,"comment":456},17,13,"该研究针对植物病害识别中的联邦学习客户端稳定化方法提出部分聚合与稳定节点方案，方法新颖、实验贴近真实场景，对智慧农业隐私计算有参考价值，值得进入每日精选。",[458],{"name":451,"url":448},[25,26,460,28,461],"联邦学习","数据隐私",{"code":38,"slug":181,"name":182,"path":183,"parent_code":184},[],{"status":186,"locked":13,"version":33,"method":34,"evidence":465},{"candidates":466,"agricultural_evidence":53},[467,472,478],{"code":38,"score":166,"supported":53,"title":468,"abstract":469,"tags":471},[],[282,283,470],"artificial intelligence",[26,28,460],{"code":44,"score":95,"supported":53,"title":473,"abstract":474,"tags":477},[],[475,476],"精准农业","precision agriculture",[25],{"code":479,"score":20,"supported":13,"title":480,"abstract":481,"tags":482},"I6",[],[],[461],[484,485],"联邦学习 植物病害识别","FedProx 作物病害检测","联邦学习植物病害识别-5192","10.3390\u002Fs26206566",{"doi":487,"openalex_id":489,"authors":490,"venue":451,"cited_by_count":61,"oa_url":448,"card":497,"direction":66,"ingested_from":153},"W7222586734",[491,494],{"name":492,"orcid":493},"Aurora-Felicia Cristea","https:\u002F\u002Forcid.org\u002F0009-0003-9718-065X",{"name":495,"orcid":496},"Ciprian Dobre","https:\u002F\u002Forcid.org\u002F0000-0003-4638-7725",{"tldr":498,"method":499,"finding":500,"direction":66,"opportunity":501},"比较联邦学习客户端级稳定化技术，提升植物病害识别的鲁棒性。","FedAvg\u002FFedAvgM\u002FFedProx聚合，仅上传骨干网络、分类头留本地，","部分聚合加稳定节点可缓解环境异质性与类别缺失带来的性能退化。","可探索面向非独立同分布农业数据的个性化联邦学习与通信效率优化。","2026-10-17T23:30:45.845684Z"]