[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-3134":3,"related-3134":71},{"id":4,"title":5,"url":6,"summary":7,"summary_zh":8,"content":9,"source_name":10,"source_url":6,"published_at":11,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":15,"score_detail":16,"sources":24,"tags":26,"search_phrases":32,"slug":35,"view_count":36,"doi":37,"paper":38,"created_at":70},3134,"DAC-Grad-CAM: Defect-aware explainable AI for automated mango quality classification","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.compag.2026.112443","Deep learning models achieve high fruit classification accuracy but lack interpretability; standard Grad-CAM produces diffuse attention without defect-specific localization. Existing explainable AI (XAI) methods for fruit grading rely on qualitative visual inspection, offering no quantitative validation of defect alignment, thereby limiting trust and regulatory acceptance in commercial quality control. This study presents an explainable mango classification framework that integrates classification performance with quantitative interpretability analysis through Defect-Aware Contrastive Grad-CAM (DAC-Grad-CAM). A new method that multiplicatively fuses gradient-based attention with an unsupervised LAB color-space bruise prior. A two-stage pipeline combining YOLO-11n detection with four classification architectures (ResNet18, MobileNetV3, VGG16, and a custom CNN). Evaluated on a three-class mango dataset (premium, class 1, class 2). Transfer-learning architectures substantially outperformed the custom CNN (ResNet18: 97.8%, VGG16: 97.6%, MobileNetV3: 96.5% vs 75.4%), a 22–23 percentage-point gap, with premium-class precision > 0.995. DAC-Grad-CAM produced the most concentrated explanations (entropy 10.02 versus 10.27 for Grad-CAM and 10.18 for SHAP). Validated against 50 manually annotated pixel-level bruise masks, it improved peak localization over standard Grad-CAM (pointing-game accuracy 0.561 versus 0.439) and, on the transfer-learning backbones, exceeded SHAP (0.659 versus 0.615) while requiring 34 ms per region against 4220 ms, a 124-fold reduction in cost. The Defect Focus Score (DFS) enabled task-specific validation, and the agreement between gradient and perturbation-based explanations (r = 0.68–0.75, p \u003C 0.001) confirmed that the localized features are task-genuine rather than method-specific artifacts. Under controlled perturbations emulating varied lighting, resolution, defocus, compression, and occlusion, defect-focused attention remained stable while classification accuracy degraded, and end-to-end throughput reached 17.4 fruit s⁻ 1 . The framework supports a confidence-based human-AI workflow (margin > 0.95, DFS > 0.37, r > 0.60) for trustworthy automated quality control in post-harvest operations.","深度学习模型虽能实现高精度水果分类，却缺乏可解释性；标准Grad-CAM产生的注意力分布弥散，无法实现缺陷特异性定位。现有用于水果分级的可解释人工智能（XAI）方法依赖定性视觉检查，无法对缺陷对齐进行定量验证，从而限制了其在商业质量控制中的可信度与法规接受度。本研究提出了一种可解释芒果分类框架，通过缺陷感知对比Grad-CAM（DAC-Grad-CAM）将分类性能与定量可解释性分析相结合。该方法将基于梯度的注意力与无监督LAB色彩空间瘀伤先验进行乘法融合。采用两阶段流水线，将YOLO-11n检测与四种分类架构（ResNet18、MobileNetV3、VGG16及自定义CNN）相结合。在三类芒果数据集（特级、一级、二级）上进行评估。迁移学习架构显著优于自定义CNN（ResNet18：97.8%，VGG16：97.6%，MobileNetV3：96.5% vs 75.4%），差距达22–23个百分点，特级类精确率> 0.995。DAC-Grad-CAM产生了最集中的解释（熵10.02，对比Grad-CAM的10.27和SHAP的10.18）。针对50个手动标注的像素级瘀伤掩膜进行验证，其峰值定位优于标准Grad-CAM（指向游戏准确率0.561 vs 0.439），且在迁移学习骨干网络上超过SHAP（0.659 vs 0.615），同时每区域仅需34 ms，而SHAP需4220 ms，成本降低124倍。缺陷聚焦分数（DFS）实现了任务特异性验证，基于梯度与基于扰动的解释之间的一致性（r = 0.68–0.75，p \u003C 0.001）证实了所定位特征是任务真实的，而非方法特异性伪影。在模拟不同光照、分辨率、散焦、压缩和遮挡的受控扰动下，缺陷聚焦注意力保持稳定，而分类准确率下降，端到端吞吐量达到17.4果·秒⁻¹。该框架支持基于置信度的人机协作工作流（margin > 0.95，DFS > 0.37，r > 0.60），可用于采后作业中可信赖的自动化质量控制。",null,"Computers and Electronics in Agriculture","2026-09-21T00:00:00Z","论文",10,false,81,{"impact":17,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":23},16,23,19,14,9,1,"方法新颖、量化验证扎实的芒果品质分级可解释AI研究，对采后智能质检有直接参考价值。",[25],{"name":10,"url":6},[27,28,29,30,31],"智慧农业","农业人工智能","可解释AI","采后处理","水果分级",[33,34],"芒果 品质分级 深度学习","DAC-Grad-CAM 芒果 缺陷检测","芒果品质分级深度学习-3134",0,"10.1016\u002Fj.compag.2026.112443",{"doi":37,"openalex_id":39,"authors":40,"venue":10,"cited_by_count":36,"oa_url":6,"card":63,"direction":67,"ingested_from":69},"W7213923596",[41,43,45,47,49,51,54,56,59,61],{"name":42,"orcid":9},"Arshed Ahmed",{"name":44,"orcid":9},"Zhao Zhang",{"name":46,"orcid":9},"Ming Li",{"name":48,"orcid":9},"Shahram Hamza Manzoor",{"name":50,"orcid":9},"Zhanjiang Zhu",{"name":52,"orcid":53},"Rashed Ahmed","https:\u002F\u002Forcid.org\u002F0009-0000-7170-2200",{"name":55,"orcid":9},"Noor Gul",{"name":57,"orcid":58},"Mustafa Mhamed","https:\u002F\u002Forcid.org\u002F0000-0002-3106-669X",{"name":60,"orcid":9},"Bing Liu",{"name":62,"orcid":9},"Mahmoud A. Abdelhamid",{"tldr":64,"method":65,"finding":66,"direction":67,"opportunity":68},"提出DAC-Grad-CAM可解释框架，实现芒果缺陷感知分类与定量验证。","融合梯度注意力与LAB色空间瘀伤先验，YOLO-11n检测加四种分类网络。","ResNet18达97.8%精度，定位优于Grad-CAM与SHAP且快124倍。","农业人工智能与决策模型","可将缺陷感知XAI扩展至多水果多缺陷，并构建人机协同质控标准。","openalex","2026-09-22T23:30:02.407104Z",{"total":72,"page":22,"page_size":72,"items":73},6,[74,119,170,229,263,297],{"id":75,"title":76,"url":77,"summary":78,"summary_zh":79,"content":9,"source_name":80,"source_url":77,"published_at":11,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":81,"score_detail":82,"sources":87,"tags":89,"search_phrases":92,"slug":95,"view_count":36,"doi":96,"paper":97,"created_at":118},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",80,{"impact":83,"substance":84,"depth":83,"authority":20,"freshness":85,"relevant":22,"comment":86},18,22,8,"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。",[88],{"name":80,"url":77},[27,28,29,90,91],"植物病害识别","多模态大模型",[93,94],"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","10.3389\u002Ffpls.2026.1937679",{"doi":96,"openalex_id":98,"authors":99,"venue":80,"cited_by_count":36,"oa_url":77,"card":112,"direction":117,"ingested_from":69},"W7213924931",[100,103,105,107,110],{"name":101,"orcid":102},"Ridong Wu","https:\u002F\u002Forcid.org\u002F0000-0003-0729-3099",{"name":104,"orcid":9},"Hua Zou",{"name":106,"orcid":9},"Zhijie Li",{"name":108,"orcid":109},"Qian Zhou","https:\u002F\u002Forcid.org\u002F0000-0001-7964-8130",{"name":111,"orcid":9},"一子 江﨑",{"tldr":113,"method":114,"finding":115,"direction":67,"opportunity":116},"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","2026-09-22T23:30:11.045394Z",{"id":120,"title":121,"url":122,"summary":123,"summary_zh":124,"content":9,"source_name":125,"source_url":122,"published_at":126,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":127,"score_detail":128,"sources":133,"tags":135,"search_phrases":138,"slug":141,"view_count":36,"doi":142,"paper":143,"created_at":169},3153,"Explainable optimized deep learning and generative AI based framework for finger millet disease detection in smart agriculture","https:\u002F\u002Fdoi.org\u002F10.1007\u002Fs44163-026-02222-y","Abstract Eleusine coracana, locally known as finger millet (ragi), is a wholesome, climate-resilient crop. It is a staple food in semi-dry and dry areas of Asia and Africa. The productivity of this crop can be severely reduced by diseases such as downy, mottle, seedling, smut, and wilt. Conventional diagnosis methods are often time-consuming and unsuitable for large-scale farming. Therefore,Deep Learning (DL) offers an alternative for smart commercial farming that can streamline disease screening and improve productivity. This work offers an AI-integrated framework for early disease intervention combining DL, Explainable AI (XAI), and an image-based decision-support interface. The finger millet (ragi) dataset from Kaggle was used to train and evaluate custom CNN, VGG16, and ResNet50 models. To reduce feature redundancy and improve generalization, Grey Wolf Optimizer (GWO) based feature selection was applied to the penultimate-layer features of each model. Custom CNN, VGG16, and ResNet50 achieved accuracies of 89.40%, 83.11%, and 91.83%, respectively. Further, the use of GWO improved the accuracies to 97.71%, 90.84%, and 98.36%, respectively. ResNet50 achieved the highest accuracy 98.36% and was selected as the final backbone. The final ResNet50 model is integrated with the Gemini API to provide disease-specific recommendations based on the predicted class and user queries. To increase prediction transparency, Grad-CAM was used to highlight image regions that influenced the model output. The integration of DL, XAI, and generative AI provides a scalable approach for detecting and managing finger millet diseases. The proposed work boosts precision agriculture by providing AI-based and XAI-supported decision-making, while fostering sustainable agricultural practices.","摘要 穇子（Eleusine coracana），当地称为指黍（ragi），是一种有益健康且气候适应性强的作物。它是亚洲和非洲半干旱及干旱地区的主食。霜霉病、斑驳病、苗枯病、黑穗病和枯萎病等病害可严重降低该作物的产量。传统诊断方法往往耗时且不适合大规模种植。因此，深度学习（DL）为智能商业农业提供了一种替代方案，可简化病害筛查并提高生产力。本研究提出了一种人工智能集成框架，用于早期病害干预，结合了深度学习、可解释人工智能（XAI）和基于图像的决策支持界面。使用来自Kaggle的指黍（ragi）数据集训练和评估了自定义CNN、VGG16和ResNet50模型。为减少特征冗余并提高泛化能力，将灰狼优化器（GWO）基于特征选择应用于每个模型的倒数第二层特征。自定义CNN、VGG16和ResNet50分别达到了89.40%、83.11%和91.83%的准确率。此外，使用GWO将准确率分别提高到97.71%、90.84%和98.36%。ResNet50达到了最高准确率98.36%，并被选为最终骨干网络。最终的ResNet50模型与Gemini API集成，根据预测类别和用户查询提供针对特定病害的建议。为提高预测透明度，使用Grad-CAM突出显示影响模型输出的图像区域。深度学习、可解释人工智能和生成式人工智能的集成提供了一种可扩展的方法，用于检测和管理指黍病害。所提出的工作通过提供基于人工智能和可解释人工智能支持的决策，促进了精准农业，同时推动了可持续农业实践。","Discover Artificial Intelligence","2026-09-22T00:00:00Z",75,{"impact":129,"substance":130,"depth":83,"authority":131,"freshness":13,"relevant":22,"comment":132},15,20,12,"将深度学习、可解释AI与生成式AI结合用于指状粟病害识别，方法新颖、数据充分，对智慧农业有参考价值。",[134],{"name":125,"url":122},[27,28,29,136,137],"病害识别","小米作物",[139,140],"ResNet50 GWO 病害识别","农业人工智能 小米作物 智慧农业 病害识别","ResNet50GWO病害识别-3153","10.1007\u002Fs44163-026-02222-y",{"doi":142,"openalex_id":144,"authors":145,"venue":125,"cited_by_count":36,"oa_url":122,"card":164,"direction":117,"ingested_from":69},"W7213978567",[146,149,151,153,155,158,161],{"name":147,"orcid":148},"Sunil Kumar Mohapatra","https:\u002F\u002Forcid.org\u002F0000-0002-5865-095X",{"name":150,"orcid":9},"A. Sanjib Kumar Patro",{"name":152,"orcid":9},"Lulen Kumar Sahu",{"name":154,"orcid":9},"Chinmaye Dora",{"name":156,"orcid":157},"Sujata Chakravarty","https:\u002F\u002Forcid.org\u002F0000-0002-1293-5378",{"name":159,"orcid":160},"Kshira Sagar Sahoo","https:\u002F\u002Forcid.org\u002F0000-0002-6435-5738",{"name":162,"orcid":163},"Byomakesh Mahapatra","https:\u002F\u002Forcid.org\u002F0000-0002-9126-1729",{"tldr":165,"method":166,"finding":167,"direction":67,"opportunity":168},"提出融合深度学习、可解释AI与生成式AI的指状粟病害检测框架，实现高精度识别与决策支持。","用Kaggle指状粟图像训练CNN、VGG16、ResNet50，并用灰狼优化做","GWO特征选择显著提升精度，ResNet50达98.36%，结合Grad-CAM与Gemini AP","可探索轻量化模型与多作物泛化，并将XAI与生成式建议在田间移动端实时验证。","2026-09-22T23:30:10.255790Z",{"id":171,"title":172,"url":173,"summary":174,"summary_zh":175,"content":9,"source_name":176,"source_url":173,"published_at":11,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":177,"score_detail":178,"sources":180,"tags":182,"search_phrases":185,"slug":188,"view_count":36,"doi":189,"paper":190,"created_at":228},3137,"From patterns to prediction: detecting crop sequence and interpreting crop selection with explainable AI at field level in Denmark","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.agsy.2026.104982","CONTEXT Crop sequences are important for sustainable agriculture, yet we know relatively little about how they recur across individual fields or which information best predicts the crop grown each year. Denmark's national field records allow both questions to be examined over time. OBJECTIVE We identified recurrent crop sequences across Denmark and tested how well annual crop-family selection could be predicted from crop history, farm characteristics, management, prices, soil and climate. METHODS Using 10 years (2011−2020) of national field-level crop data (∼600,000 fields yr −1 ), we developed a heuristic algorithm to detect recurring crop sequence patterns. We then trained machine learning (LightGBM) and deep learning (TabNet) models to predict annual crop choices based on preceding crops and lagged management and farm structure predictors, and pedoclimatic conditions. Models were evaluated through forward-chaining validation and temporal holdout tests, and SHAP values were used to examine how LightGBM used each predictor. RESULTS AND DISCUSSION Crop sequences were largely dominated by cereals, with little diversification even in longer sequences. Crop selection patterns were strongly associated with the previous crops and farm typology, but pedoclimatic conditions and previous management played a minor role. The DL model achieved a slightly higher overall accuracy, particularly for dominant crops, while the ML model provided a more balanced performance across different crops and enabled interpretation through SHAP. SIGNIFICANCE The analysis separates two distinct tasks: identifying multi-year crop sequences and predicting the crop family grown each year. The sequence analysis shows why diversification cannot be assessed from sequence length or crop counts alone. The predictive models could help generate locally plausible sequences for agri-environmental simulations, reducing reliance on standard rotations that poorly reflect observed field histories.","背景 作物序列对可持续农业至关重要，但我们对它们在单个田块中如何重复出现，以及哪些信息最能预测每年种植的作物知之甚少。丹麦的国家田块记录使这两个问题得以在长时间尺度上加以考察。 目标 我们识别了丹麦全国范围内重复出现的作物序列，并检验了年度作物科选择能在多大程度上由作物历史、农场特征、管理、价格、土壤和气候来预测。 方法 利用10年（2011−2020年）的国家田块级作物数据（每年约60万块田），我们开发了一种启发式算法来检测重复出现的作物序列模式。随后，我们训练了机器学习（LightGBM）和深度学习（TabNet）模型，基于前茬作物以及滞后的管理和农场结构预测因子及土壤气候条件来预测年度作物选择。模型通过前向链式验证和时间留出测试进行评估，并使用SHAP值考察LightGBM如何利用每个预测因子。 结果与讨论 作物序列在很大程度上以谷类作物为主，即使在较长的序列中也几乎没有多样化。作物选择模式与前茬作物和农场类型密切相关，而土壤气候条件和先前管理的作用较小。深度学习模型取得了略高的总体准确率，尤其是在优势作物上，而机器学习模型在不同作物之间提供了更均衡的表现，并可通过SHAP进行解释。 意义 该分析区分了两个不同的任务：识别多年作物序列和预测每年种植的作物科。序列分析表明，为什么不能仅凭序列长度或作物计数来评估多样化。预测模型有助于为农业环境模拟生成局部合理的序列，减少对不能很好反映观测田块历史的标准轮作方式的依赖。","Agricultural Systems",79,{"impact":17,"substance":84,"depth":83,"authority":20,"freshness":21,"relevant":22,"comment":179},"基于丹麦十年全国田块数据，用可解释AI预测作物选择，方法新颖、数据规模大，对农业信息化与轮作模拟有参考价值。",[181],{"name":176,"url":173},[27,28,29,183,184],"作物轮作","丹麦农业",[186,187],"丹麦 作物序列 机器学习","LightGBM 作物选择 预测","丹麦作物序列机器学习-3137","10.1016\u002Fj.agsy.2026.104982",{"doi":189,"openalex_id":191,"authors":192,"venue":176,"cited_by_count":36,"oa_url":173,"card":223,"direction":67,"ingested_from":69},"W7213919617",[193,195,198,201,203,205,207,209,212,214,217,220],{"name":194,"orcid":9},"João G. Serra",{"name":196,"orcid":197},"Edwin Haas","https:\u002F\u002Forcid.org\u002F0000-0003-0664-642X",{"name":199,"orcid":200},"Diego Ábalos","https:\u002F\u002Forcid.org\u002F0000-0002-4189-5563",{"name":202,"orcid":9},"Søren Kolind Hvid",{"name":204,"orcid":9},"Tommy Dalgaard",{"name":206,"orcid":9},"Lars Uldall-Jessen",{"name":208,"orcid":9},"Franca Giannini-Kurina",{"name":210,"orcid":211},"Meshach Ojo Aderele","https:\u002F\u002Forcid.org\u002F0009-0007-9381-6445",{"name":213,"orcid":9},"Bo Thiesson",{"name":215,"orcid":216},"Jørgen Eivind Olesen","https:\u002F\u002Forcid.org\u002F0000-0002-6639-1273",{"name":218,"orcid":219},"Klaus Butterbach‐Bahl","https:\u002F\u002Forcid.org\u002F0000-0001-9499-6598",{"name":221,"orcid":222},"Jaber Rahimi","https:\u002F\u002Forcid.org\u002F0000-0002-2754-2358",{"tldr":224,"method":225,"finding":226,"direction":67,"opportunity":227},"用丹麦十年田块数据识别轮作模式并预测年度作物选择。","60万田块年数据，启发式序列检测+LightGBM\u002FTabNet，SHAP解释。","作物序列以谷物为主、多样化低；作物选择主要由前茬和农场类型决定。","可结合可解释AI生成区域化可行轮作方案，替代脱离实际的标准化轮作假设。","2026-09-22T23:30:05.518307Z",{"id":230,"title":231,"url":232,"summary":233,"summary_zh":234,"content":9,"source_name":235,"source_url":232,"published_at":236,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":237,"score_detail":238,"sources":242,"tags":244,"search_phrases":247,"slug":250,"view_count":36,"doi":251,"paper":252,"created_at":262},2518,"AI-Driven Transformation in Sustainable Agriculture: A Systematic Review","https:\u002F\u002Fdoi.org\u002F10.18805\u002Fijare.a-6627","The agriculture sector is facing unprecedented pressure from a rapidly growing global population, erratic weather conditions and declining water availability. Traditional farming methods are increasingly proving insufficient in meeting the rising food demand, necessitating a shift towards technology-driven solutions. This paper provides a critical and systematic review of the application of Artificial Intelligence (AI) in agriculture, analysing its impact on productivity and sustainability, while identifying potential barriers hindering adoption. Employing a Systematic Literature Review approach, this study synthesizes recent research organized around the core domains of agricultural production - crop, soil and nutrient, water and irrigation and crop-protection (pest, disease and weed) management, alongside yield forecasting - examining how data-driven tools such as machine learning, the Internet of Things, robotics and computer vision are applied within each and prioritizing studies that reported agronomically meaningful outcomes such as yield gains and resource-use efficiency. The strongest gains identified across the review fall within three agricultural areas - crop and soil monitoring, yield forecasting and robotic field operations such as targeted detection, weeding and harvesting. The technologies enabling them, chiefly machine-learning models such as convolutional neural networks combined with field and remote sensors, have delivered high in-field precision, with reported object-detection precision of up to 95.78% in field conditions. In practical terms, this translates into more efficient use of water, fertilizer and other inputs and into improved yield and crop quality. The review concludes that prioritizing Explainable AI and interoperable farm systems is central to building farmer trust and safeguarding food security under a changing climate.","农业部门正面临着来自全球人口快速增长、天气条件不稳定以及水资源日益减少的前所未有的压力。传统耕作方法在满足不断增长的粮食需求方面日益显得力不从心，因此有必要转向技术驱动的解决方案。本文对人工智能（Artificial Intelligence, AI）在农业中的应用进行了批判性和系统性的综述，分析了其对生产力和可持续性的影响，同时识别了阻碍其采用的潜在障碍。本研究采用系统性文献综述方法，围绕农业生产的核心领域——作物、土壤与养分、水资源与灌溉、作物保护（病虫害和杂草）管理以及产量预测——对近期研究进行了综合梳理，考察了机器学习、物联网、机器人技术和计算机视觉等数据驱动工具在各领域中的应用方式，并优先关注报告了具有农学意义成果（如产量提升和资源利用效率）的研究。综述中发现的最显著收益集中在三个农业领域——作物与土壤监测、产量预测以及机器人田间作业（如靶向检测、除草和收获）。实现这些收益的技术，主要是卷积神经网络等机器学习模型与田间及遥感传感器的结合，已在田间条件下实现了高精度，所报告的目标检测精度在田间条件下高达95.78%。在实际应用中，这意味着水、肥料及其他投入品的更高效利用，以及产量和作物品质的提升。综述得出结论：优先发展可解释人工智能（Explainable AI）和可互操作的农场系统，对于在气候变化背景下建立农民信任和保障粮食安全至关重要。","Indian Journal of Agricultural Research","2026-09-14T00:00:00Z",76,{"impact":83,"substance":130,"depth":239,"authority":240,"freshness":85,"relevant":22,"comment":241},17,13,"系统综述梳理AI在作物土壤监测、产量预测与机器人田间作业中的应用成效与推广障碍，结论扎实，对智慧农业方向有参考价值。",[243],{"name":235,"url":232},[27,28,245,29,246],"产量预测","精准农业",[248,249],"农业人工智能 产量预测 智慧农业 精准农业","农业人工智能 产量预测","农业人工智能产量预测智慧农业精准农业-2518","10.18805\u002Fijare.a-6627",{"doi":251,"openalex_id":253,"authors":254,"venue":235,"cited_by_count":36,"oa_url":232,"card":257,"direction":117,"ingested_from":69},"W7212616413",[255],{"name":256,"orcid":9},"Manish Maan",{"tldr":258,"method":259,"finding":260,"direction":67,"opportunity":261},"系统综述AI在可持续农业中的应用，分析生产力与可持续性影响及推广障碍。","系统文献综述，围绕作物、土壤、水、植保和产量预测梳理ML、IoT、机器人、计算机","作物与土壤监测、产量预测和机器人田间作业收益最大，检测精度达95.78%，需可解释AI与互操作系统。","可解释AI与互操作农场系统在农户信任和气候适应中的作用尚缺实证，是值得深入的研究空白。","2026-09-15T23:30:13.705421Z",{"id":264,"title":265,"url":266,"summary":267,"summary_zh":268,"content":9,"source_name":269,"source_url":266,"published_at":236,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":270,"score_detail":271,"sources":273,"tags":275,"search_phrases":278,"slug":281,"view_count":36,"doi":282,"paper":283,"created_at":296},2517,"Artificial Intelligence-Based Intelligent Fertigation Recommendation Systems for Precision Agriculture","https:\u002F\u002Fdoi.org\u002F10.64388\u002Firev10i3-1722972","Artificial-intelligence-enabled fertilizer and fertigation recommendation has progressed from single-task crop or fertilizer classification toward integrated decision-support architectures. However, predictive classification, agronomic dose calculation, irrigation scheduling, explainability, real-time sensing, and farmer-facing delivery are often studied separately. This paper presents a structured narrative review of these strands and an applied machine-learning case study using the publicly available Crop and Fertilizer Dataset for Western Maharashtra. The case study contains 4,513 records spanning five districts, 16 crops, and 19 fertilizer classes. Seven classifiers were compared using an 80:20 stratified train-test split with five-fold stratified cross-validation on the training set. XGBoost achieved 97.34% test accuracy and 94.21% ± 1.15% cross-validated accuracy, while Random Forest achieved 93.58% and 90.89% ± 0.80%, respectively. TreeSHAP analysis of Random Forest identified crop identity, potassium, and nitrogen as the leading predictors of the historical fertilizer class. These results are interpreted as a computational baseline rather than proof of agronomic optimality because the target label represents recorded fertilizer choices. The review also incorporates evidence on IoT\u002Fedge-cloud sensing, multilingual agricultural advisory, and federated learning. It concludes that RF\u002FSHAP, IoT sensing, and multilingual interfaces are established capabilities; a more defensible research direction is an auditable pipeline that separates fertilizer identity, nutrient dose, and application timing, connects explainable prediction to sequential scheduling, and independently benchmarks outputs against authoritative agronomic guidance. The proposed Intelligent Fertigation Recommendation System (IFRS) is therefore presented as a research framework requiring multi-season and field validation before claims of yield, water, nutrient-use-efficiency, or adoption benefits.","人工智能驱动的肥料与灌溉施肥推荐已从单一任务的作物或肥料分类，发展为集成式决策支持架构。然而，预测性分类、农艺用量计算、灌溉调度、可解释性、实时感知以及面向农户的交付往往被分别研究。本文对这些研究脉络进行了结构化叙述性综述，并基于公开的“西马哈拉施特拉邦作物与肥料数据集”开展了一项应用机器学习案例研究。该案例研究包含4，513条记录，涵盖五个地区、16种作物和19个肥料类别。采用80：20分层训练-测试划分，并在训练集上进行五折分层交叉验证，比较了七种分类器。XGBoost取得了97.34%的测试准确率和94.21% ± 1.15%的交叉验证准确率，而随机森林分别取得了93.58%和90.89% ± 0.80%。对随机森林的TreeSHAP分析表明，作物身份、钾和氮是历史肥料类别的主要预测因子。这些结果被解释为计算基线，而非农艺最优性的证明，因为目标标签代表的是有记录的肥料选择。该综述还纳入了关于物联网\u002F边缘-云感知、多语言农业咨询和联邦学习的证据。结论认为，随机森林\u002FSHAP、物联网感知和多语言界面已是成熟能力；更具可辩护性的研究方向是构建一条可审计的流水线，将肥料身份、养分用量和施用时机分离，将可解释预测与序贯调度相连接，并依据权威农艺指南对输出进行独立基准测试。因此，所提出的智能灌溉施肥推荐系统（IFRS）被作为一个研究框架提出，在声称产量、水分、养分利用效率或采用效益之前，仍需进行多季和田间验证。","Iconic Research and Engineering Journals",62,{"impact":131,"substance":83,"depth":239,"authority":72,"freshness":21,"relevant":22,"comment":272},"对AI水肥推荐研究进行系统综述并给出可复现的机器学习基线，方法透明、结论审慎，但来源为普通工程类期刊且属综述性论文，产业影响有限，适合作为技术参考而非每日精选头条。",[274],{"name":269,"url":266},[27,28,276,29,277],"精准施肥","水肥一体化",[279,280],"农业人工智能 水肥一体化 智慧农业 精准施肥","农业人工智能 水肥一体化","农业人工智能水肥一体化智慧农业精准施肥-2517","10.64388\u002Firev10i3-1722972",{"doi":282,"openalex_id":284,"authors":285,"venue":269,"cited_by_count":36,"oa_url":290,"card":291,"direction":117,"ingested_from":69},"W7212934364",[286,288],{"name":287,"orcid":9},"Shraddha S. Tayade",{"name":289,"orcid":9},"Yogesh V. Chimate","https:\u002F\u002Fwww.irejournals.com\u002Fformatedpaper\u002F1722972.pdf",{"tldr":292,"method":293,"finding":294,"direction":67,"opportunity":295},"综述AI施肥推荐研究，并用马哈拉施特拉数据集比较七种分类器，提出可审计智能施肥推荐框架。","结构化综述加机器学习案例，4513条记录，七分类器对比，XGBoost与Tree","XGBoost测试准确率97.34%，但标签仅为历史施肥选择，不能证明农艺最优。","可研究分离肥料种类、养分剂量与施用时序的可审计推荐流水线，并进行多季田间验证。","2026-09-15T23:30:12.675690Z",{"id":298,"title":299,"url":300,"summary":301,"summary_zh":302,"content":9,"source_name":10,"source_url":300,"published_at":236,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":303,"score_detail":304,"sources":306,"tags":308,"search_phrases":311,"slug":314,"view_count":36,"doi":315,"paper":316,"created_at":340},2499,"Real-time on-tree Korla pear grading with a shared ordinal–anomaly descriptor field","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.compag.2026.112371","On-tree grading of Korla fragrant pears couples two decisions of different nature: a gradual Grade-A\u002FGrade-B appearance transition and a Grade-C surface-defect judgement, both made under illumination change, occlusion and scale variation. Detectors that treat the fruit as a single class, and graders that place the three grades on one flat classification axis, optimise a small A\u002FB deviation and an A\u002FC defect error through the same geometry and leave no explicit account of what ordinal and defect information each spatial location should preserve. We address this by adding an explicit ordinal–anomaly descriptor field to the shared feature hierarchy. A shallow module generates three channels – ordinal tendency, defect tendency and foreground confidence – that condition four downstream detector stages, and a hierarchical head converts the two kinds of evidence into normalised grade probabilities. On the orchard-disjoint KPR-3 dataset (3826 images, 12,458 instances, three commercial orchards), the model reaches 94.0% mAP@50 and 92.6% matched-instance grade accuracy, and lowers the unsafe-pick rate from 4.8% to 2.1%. On a Jetson Orin Nano at 15 W, the INT8 engine runs at 93.1% mAP@50 and sustains 42 FPS of end-to-end pipelined throughput. Monocular sensing remains a key limitation under severe front–back fruit overlap.","库尔勒香梨的树上分级耦合了两个性质不同的决策：渐变的A级\u002FB级外观过渡，以及C级表面缺陷判定，二者均在光照变化、遮挡和尺度变化下做出。将果实视为单一类别的检测器，以及将三个等级置于同一扁平分类轴上的分级器，通过相同的几何结构优化较小的A\u002FB偏差和A\u002FC缺陷误差，却未明确说明每个空间位置应保留何种序数信息和缺陷信息。我们通过在共享特征层级中增加一个显式的序数–异常描述符场来解决这一问题。一个浅层模块生成三个通道——序数倾向、缺陷倾向和前景置信度——用以调节四个下游检测器阶段，一个层级式头部将两类证据转换为归一化等级概率。在果园不相交的KPR-3数据集（3826幅图像、12458个实例、三个商业果园）上，模型达到94.0% mAP@50和92.6%的匹配实例等级准确率，并将不安全采摘率从4.8%降至2.1%。在15 W的Jetson Orin Nano上，INT8引擎以93.1% mAP@50运行，并维持42 FPS的端到端流水线吞吐量。在严重前后果实重叠情况下，单目感知仍是主要局限。",83,{"impact":83,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":305},"提出序数-异常描述子场的树上香梨实时分级方法，在自建KPR-3数据集与Jetson边缘端验证，方法新颖、数据扎实，对果园采摘机器人有直接参考价值。",[307],{"name":10,"url":300},[27,28,309,310,31],"边缘计算","库尔勒香梨",[312,313],"农业人工智能 库尔勒香梨 智慧农业 水果分级","农业人工智能 库尔勒香梨","农业人工智能库尔勒香梨智慧农业水果分级-2499","10.1016\u002Fj.compag.2026.112371",{"doi":315,"openalex_id":317,"authors":318,"venue":10,"cited_by_count":36,"oa_url":300,"card":335,"direction":117,"ingested_from":69},"W7212908812",[319,321,324,326,328,330,332],{"name":320,"orcid":9},"Bingyu Cao",{"name":322,"orcid":323},"Peng Zhou","https:\u002F\u002Forcid.org\u002F0000-0002-6345-5307",{"name":325,"orcid":9},"Zhikai Yang",{"name":327,"orcid":9},"Wei Chen",{"name":329,"orcid":9},"Yingchao Wang",{"name":331,"orcid":9},"Mingqi Kan",{"name":333,"orcid":334},"Haiyong Chen","https:\u002F\u002Forcid.org\u002F0000-0002-5262-4208",{"tldr":336,"method":337,"finding":338,"direction":117,"opportunity":339},"提出共享序数-异常描述子场，实现树上库尔勒香梨实时分级。","在共享特征层级加入序数、缺陷、前景三通道描述子，用KPR-3数据集训练。","mAP@50达94.0%，分级准确率92.6%，不安全采摘率从4.8%降至2.1%。","可探索多模态或深度传感融合，解决严重前后遮挡下的单目感知局限。","2026-09-15T23:30:01.440708Z"]