[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-1597":3,"related-1597":88},{"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":23,"tags":25,"primary_topic":9,"secondary_topics":30,"classification":31,"search_phrases":59,"slug":62,"view_count":20,"doi":63,"paper":64,"created_at":87},1597,"Lite-HCAR-CNN: A Lightweight Deep Learning Model for Rapid Forest Soil Organic Carbon Estimation from Laboratory Vis-NIR Spectra","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.atech.2026.102530","Rapid and reliable monitoring of soil organic carbon (SOC) is important for precision forest management and soil ecological assessment. In this study, 853 surface soil samples were collected from subtropical plantation and secondary forest regions in Guangxi, China. After spectral quality control and SOC outlier screening, 842 valid samples were retained for model development. Laboratory-acquired visible and near-infrared spectra were preprocessed using Savitzky-Golay smoothing and standard normal variate correction, followed by a dimensionality-reduction comparison involving the full spectrum, 10 nm, 20 nm and 40 nm binning, principal component analysis, and PLS-VIP feature selection. To improve computational efficiency for rapid SOC estimation from Vis-NIR spectra, a lightweight model named Lite-HCAR-CNN was developed. The model adopts two one-dimensional convolutional blocks with a kernel size of 7, dual squeeze-and-excitation channel recalibration, Swish activation, and a compact residual dense regression head. Model evaluation employed an SOC-stratified outer five-fold cross-validation strategy, with an inner validation subset used only for checkpoint selection and each held-out outer test fold evaluated once. Under the selected 20 nm binning strategy, Lite-HCAR-CNN achieved an average R² of 0.810 ± 0.027 and RMSE of 4.617 ± 0.373 g\u002Fkg. Compared with 1D-CNN, Lite-HCAR-CNN reduced RMSE by approximately 8.4%, and compared with 1D-ResNet18 it reduced trainable parameters by 87.9%, model storage size by 87.9%, FLOPs by 96.7%, and CPU inference latency by 88.6%. Binning comparison, kernel-size and hidden-unit sensitivity analyses, ablation experiments, residual diagnosis, and MC Dropout uncertainty analysis were further conducted to clarify the model design and reliability. The results indicate that Lite-HCAR-CNN achieves a favorable balance between prediction accuracy and computational efficiency for fast estimation of forest soil organic carbon using laboratory Vis-NIR spectroscopy.","土壤有机碳（SOC）的快速可靠监测对于精准森林管理和土壤生态评估具有重要意义。本研究从中国广西的亚热带人工林和次生林区域采集了853份表层土壤样品，经光谱质量控制和SOC异常值筛选后，保留842份有效样品用于模型构建。实验室获取的可见-近红外光谱采用Savitzky-Golay平滑和标准正态变量校正进行预处理，随后对全光谱、10 nm、20 nm和40 nm分箱、主成分分析及PLS-VIP特征选择进行了降维比较。为提高基于可见-近红外光谱快速估算SOC的计算效率，开发了一种轻量级模型Lite-HCAR-CNN。该模型采用两个核大小为7的一维卷积块、双重挤压-激励通道重标定、Swish激活函数以及紧凑的残差密集回归头。模型评估采用基于SOC分层的五折外交叉验证策略，内验证子集仅用于检查点选择，每个保留的外测试折仅评估一次。在选定的20 nm分箱策略下，Lite-HCAR-CNN实现了平均R²为0.810 ± 0.027，RMSE为4.617 ± 0.373 g\u002Fkg。与1D-CNN相比，Lite-HCAR-CNN的RMSE降低了约8.4%；与1D-ResNet18相比，可训练参数减少了87.9%，模型存储大小减少了87.9%，FLOPs减少了96.7%，CPU推理延迟降低了88.6%。进一步开展了分箱比较、核大小和隐藏单元敏感性分析、消融实验、残差诊断及MC Dropout不确定性分析，以阐明模型设计及其可靠性。结果表明，Lite-HCAR-CNN在利用实验室可见-近红外光谱快速估算森林土壤有机碳方面，实现了预测精度与计算效率之间的良好平衡。",null,"Smart Agricultural Technology","2026-09-01T00:00:00Z","论文",10,false,68,{"impact":17,"substance":18,"depth":19,"authority":17,"freshness":20,"relevant":21,"comment":22},12,22,18,4,1,"提出轻量化深度学习模型，兼顾精度与效率，对森林土壤碳估算有实用价值。",[24],{"name":10,"url":6},[26,27,28,29],"智慧农业","农业人工智能","土壤监测","模型优化",[],{"status":32,"locked":14,"version":33,"method":34,"evidence":35},"needs_review","2026-10-22.2","rules",{"candidates":36,"agricultural_evidence":14,"reason":58},[37,46,53],{"code":38,"score":39,"supported":40,"title":41,"abstract":43,"tags":45},"H2",14,true,[42],"soil",[44,42],"土壤",[],{"code":47,"score":48,"supported":40,"title":49,"abstract":51,"tags":52},"I2",9,[50],"deep learning",[],[27],{"code":54,"score":21,"supported":14,"title":55,"abstract":56,"tags":57},"I1",[],[],[26],"农业相关性需要复核",[60,61],"农业人工智能 土壤监测 智慧农业 模型优化","农业人工智能 土壤监测","农业人工智能土壤监测智慧农业模型优化-1597","10.1016\u002Fj.atech.2026.102530",{"doi":63,"openalex_id":65,"authors":66,"venue":10,"cited_by_count":78,"oa_url":79,"card":80,"direction":84,"ingested_from":86},"W7207551174",[67,69,72,74,76],{"name":68,"orcid":9},"Jian Tang",{"name":70,"orcid":71},"Zubo Meng","https:\u002F\u002Forcid.org\u002F0009-0008-9632-5067",{"name":73,"orcid":9},"Yuanyuan Shi",{"name":75,"orcid":9},"Yun Deng",{"name":77,"orcid":9},"Junyu Zhao",0,"https:\u002F\u002Fwww.sciencedirect.com\u002Fscience\u002Farticle\u002Fpii\u002FS2772375526007550\u002Fpdf",{"tldr":81,"method":82,"finding":83,"direction":84,"opportunity":85},"提出轻量级深度学习模型Lite-HCAR-CNN，用于从实验室可见-近红外光谱快速估算森林土壤有机碳","采用一维卷积、挤压激励、Swish激活及残差密集回归头，结合20nm分箱和SG-","模型在精度与效率间取得平衡，RMSE降低8.4%，参数减少87.9%，推理延迟降低88.6%。","农业遥感与作物表型","可探索将轻量级模型部署于便携式或机载光谱设备，实现野外实时土壤碳监测，并扩展至其他土壤属性。","openalex","2026-09-04T23:30:04.459941Z",{"total":89,"page":21,"page_size":89,"items":90},6,[91,189,245,340,414,475],{"id":92,"title":93,"url":94,"summary":95,"summary_zh":96,"content":9,"source_name":97,"source_url":94,"published_at":98,"category":12,"cover_url":9,"hotness":13,"is_selected":40,"score":99,"score_detail":100,"sources":102,"tags":104,"primary_topic":107,"secondary_topics":111,"classification":118,"search_phrases":152,"slug":155,"view_count":21,"doi":156,"paper":157,"created_at":188},4345,"HyBEAR: a hyperspectral benchmark for bare soil detection","https:\u002F\u002Fdoi.org\u002F10.5194\u002Fessd-18-7301-2026","Detecting bare soil areas is an important step in the analysis of Earth observation data in a variety of Precision Agriculture (PA) applications focused on quantifying soil properties and assessing soil quality. In this paper, we introduce the HyBEAR benchmark – a novel large-scale collection of high-resolution hyperspectral aerial images (with 2 m ground sampling distance) accompanied by manual bare soil annotations verified by domain experts. Usually, the bare soil detection problem is tackled at the pixel level, meaning that detection methods classify all pixels as either bare soil or background. In contrast to this approach, we provide pixel-level annotations for the entire agricultural parcels (if the parcel is labeled as bare soil, then all pixels within that parcel are labeled accordingly), and aim to support the development of methods that identify entire fields with no vegetation. Commonly, such fields undergo further analysis to determine specific soil parameters and characteristics that are important when planning various PA activities, such as fertilization. The HyBEAR benchmark includes (i) the largest-to-date (108 064 591 pixels, corresponding to 43 225 ha) and most heterogeneous dataset for bare soil detection, as well as (ii) the validation procedure (training-test splits and quality metrics) and a set of baseline results, obtained for a set of machine learning bare soil detection models. From the FULL collection of 1954 images in HyBEAR, which we divided into 5 spatially-disjoint folds, we additionally selected a random, stratified subset (MINI) of the images which may be useful for designing and verifying bare soil detection algorithms. Overall, HyBEAR is a step toward standardizing the way the community builds and confronts bare soil detection algorithms in a thorough, reproducible, and unbiased way. The dataset is publicly available at Zenodo: https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.17607897 (Wijata et al., 2025).","在多种精准农业（PA）应用中，裸土区域检测是对地观测数据分析的重要步骤，其重点在于量化土壤性质和评估土壤质量。本文介绍了HyBEAR基准——一个全新的大规模高分辨率高光谱航空影像数据集（地面采样距离为2 m），并附有经领域专家验证的人工裸土标注。通常，裸土检测问题在像素层面进行处理，即检测方法将所有像素分类为裸土或背景。与此方法不同，我们为整个农业地块提供像素级标注（如果地块被标记为裸土，则该地块内的所有像素均被相应标记），旨在支持开发能够识别无植被整块田地的方法。通常，此类田地会经过进一步分析以确定特定的土壤参数和特征，这些在规划施肥等各种精准农业活动时至关重要。HyBEAR基准包含：（i）迄今为止规模最大（108 064 591个像素，对应43 225公顷）且最具异质性的裸土检测数据集；（ii）验证流程（训练-测试划分和质量指标）以及一组基线结果，这些结果是在一组机器学习裸土检测模型上获得的。从HyBEAR中全部1954幅图像（我们将其划分为5个空间不相交的折）中，我们额外选择了一个随机分层子集（MINI），该子集可能有助于设计和验证裸土检测算法。总体而言，HyBEAR朝着以全面、可复现且无偏见的方式标准化社区构建和对比裸土检测算法的方向迈出了一步。该数据集已在Zenodo上公开：https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.17607897（Wijata et al., 2025）。","Earth system science data","2026-10-06T00:00:00Z",81,{"impact":19,"substance":18,"depth":19,"authority":39,"freshness":48,"relevant":21,"comment":101},"发布迄今最大规模高光谱裸土检测基准数据集，含像素级标注与基线结果，对精准农业土壤分析有实质推动。",[103],{"name":97,"url":94},[26,27,105,106,28],"高光谱","遥感",{"code":38,"slug":42,"name":108,"path":109,"parent_code":110},"土壤健康与改良","\u002Ftopic\u002Fenvironment\u002Fsoil","H",[112],{"code":113,"slug":114,"name":115,"path":116,"parent_code":117},"I3","remote-sensing","遥感与农情监测","\u002Ftopic\u002Fdigital-agriculture\u002Fremote-sensing","I",{"status":119,"locked":14,"version":33,"method":34,"evidence":120},"classified",{"candidates":121,"agricultural_evidence":40},[122,126,132,139,145],{"code":38,"score":39,"supported":40,"title":123,"abstract":124,"tags":125},[42],[44,42],[],{"code":113,"score":127,"supported":40,"title":128,"abstract":130,"tags":131},13,[129],"hyperspectral",[129],[106,105],{"code":54,"score":133,"supported":40,"title":134,"abstract":135,"tags":138},7,[],[136,137],"精准农业","precision agriculture",[26],{"code":47,"score":133,"supported":40,"title":140,"abstract":141,"tags":144},[],[142,143],"机器学习","machine learning",[27],{"code":146,"score":89,"supported":40,"title":147,"abstract":148,"tags":151},"F3",[],[149,150],"施肥","fertilization",[],[153,154],"HyBEAR 裸土检测 高光谱","裸土检测 基准数据集 遥感","HyBEAR裸土检测高光谱-4345","10.5194\u002Fessd-18-7301-2026",{"doi":156,"openalex_id":158,"authors":159,"venue":97,"cited_by_count":78,"oa_url":181,"card":182,"direction":187,"ingested_from":86},"W7220738771",[160,163,166,169,172,175,178],{"name":161,"orcid":162},"Agata M. Wijata","https:\u002F\u002Forcid.org\u002F0000-0001-6180-9979",{"name":164,"orcid":165},"Bogdan Ruszczak","https:\u002F\u002Forcid.org\u002F0000-0003-1089-1778",{"name":167,"orcid":168},"Adriana Niepala","https:\u002F\u002Forcid.org\u002F0009-0005-8274-9852",{"name":170,"orcid":171},"Michał Gumiela","https:\u002F\u002Forcid.org\u002F0009-0000-2501-2786",{"name":173,"orcid":174},"Krzysztof Smykała","https:\u002F\u002Forcid.org\u002F0000-0003-1970-5388",{"name":176,"orcid":177},"Nicolas Longépé","https:\u002F\u002Forcid.org\u002F0000-0002-6832-3274",{"name":179,"orcid":180},"Jakub Nalepa","https:\u002F\u002Forcid.org\u002F0000-0002-4026-1569","https:\u002F\u002Fessd.copernicus.org\u002Farticles\u002F18\u002F7301\u002F2026\u002Fessd-18-7301-2026.pdf",{"tldr":183,"method":184,"finding":185,"direction":84,"opportunity":186},"提出HyBEAR高光谱航空影像基准数据集，用于整田块裸土检测。","2米分辨率高光谱航空影像，专家验证的田块级像素标注，含训练测试划分与基线模型。","构建了迄今最大（43225公顷）且异质的裸土检测数据集，并给出可复现的评估流程与基线结果。","可基于该基准探索田块级裸土检测的迁移学习与多光谱\u002F高光谱融合方法，并关联土壤参数反演。","智慧农业 \u002F 农业物联网","2026-10-07T23:30:09.147409Z",{"id":190,"title":191,"url":192,"summary":193,"summary_zh":9,"content":9,"source_name":194,"source_url":9,"published_at":195,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":196,"score_detail":197,"sources":199,"tags":201,"primary_topic":202,"secondary_topics":206,"classification":208,"search_phrases":231,"slug":234,"view_count":20,"doi":9,"paper":235,"created_at":244},4286,"Applications of Artificial Intelligence in Soil Characterization and Agriculture: A Systematic Review","https:\u002F\u002Fm2.mtmt.hu\u002Fapi\u002Fpublication\u002F37386408","Navarro Rubio等系统综述2018-2026年196项AI土壤特性与农业应用研究，遵循PRISMA 2020框架。综述发现研究明显向数据驱动方法转变，机器学习和深度学习模型主导近期研究；随机森林、支持向量机、梯度提升方法、人工神经网络、卷积神经网络、长短期记忆架构最常被报告；主要数据源包括原位传感、实验室测量、遥感影像和环境协变量，常通过多源数据融合框架集成；树集成模型在不同土壤特性上提供稳健性能。","Agronomy 2026 16(13):1241 DOI:10.3390\u002Fagronomy16131241","2026-10-02T00:00:00Z",78,{"impact":19,"substance":18,"depth":19,"authority":39,"freshness":89,"relevant":21,"comment":198},"PRISMA框架下196项研究的系统综述，方法学扎实、结论可靠，对农业AI土壤应用具较高参考价值。",[200],{"name":194,"url":192},[26,27,142,106,28],{"code":47,"slug":203,"name":204,"path":205,"parent_code":117},"agricultural-ai","农业AI与大模型","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-ai",[207],{"code":38,"slug":42,"name":108,"path":109,"parent_code":110},{"status":119,"locked":14,"version":33,"method":34,"evidence":209},{"candidates":210,"agricultural_evidence":40},[211,218,223,227],{"code":47,"score":212,"supported":40,"title":213,"abstract":215,"tags":217},16,[214],"artificial intelligence",[216,142],"深度学习",[27,142],{"code":38,"score":219,"supported":40,"title":220,"abstract":221,"tags":222},11,[42],[44],[],{"code":113,"score":20,"supported":40,"title":224,"abstract":225,"tags":226},[],[106],[106],{"code":54,"score":21,"supported":14,"title":228,"abstract":229,"tags":230},[],[],[26],[232,233],"AI 土壤特性 系统综述","机器学习 土壤 遥感","AI土壤特性系统综述-4286",{"doi":9,"openalex_id":9,"authors":236,"venue":9,"cited_by_count":78,"oa_url":9,"card":237,"direction":241,"ingested_from":243},[],{"tldr":238,"method":239,"finding":240,"direction":241,"opportunity":242},"系统综述2018-2026年196项AI用于土壤特性与农业的研究。","PRISMA 2020框架，分析机器学习与深度学习模型及多源数据融合。","研究转向数据驱动，树集成模型在不同土壤特性上表现稳健。","农业人工智能与决策模型","可探索可解释AI与多源数据融合在土壤特性预测中的泛化能力。","agent","2026-10-06T00:05:04.976398Z",{"id":246,"title":247,"url":248,"summary":249,"summary_zh":250,"content":9,"source_name":251,"source_url":248,"published_at":252,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":253,"score_detail":254,"sources":258,"tags":260,"primary_topic":262,"secondary_topics":267,"classification":274,"search_phrases":322,"slug":325,"view_count":21,"doi":326,"paper":327,"created_at":339},4073,"Intelligent soil monitoring system for sustainable agriculture using TinyML and ultra-low power wireless sensor networks","https:\u002F\u002Fdoi.org\u002F10.35784\u002Fiapgos.9467","Precision agriculture demands continuous, energy-autonomous soil monitoring capable of detecting critical conditions such as rapid moisture loss or nutrient depletion approaching crop-threatening thresholds. Conventional wireless sensor networks face a fundamental trade-off between multi-year battery life and sub-second response latency, as duty-cycled polling wastes energy on idle queries while always-on communication remains infeasible for battery-powered field deployments spanning hundreds of acres. This work presents a novel integration of TinyML-based edge inference with a custom-designed 40 nm CMOS ultra-low power wake-up receiver (WuRX), enabling event-driven, asynchronous soil condition monitoring without reliance on periodic polling schedules. The proposed three-tier edge-fog-cloud architecture employs sensor nodes measuring moisture, temperature, pH, nitrogen, phosphorus, and potassium, with quantized neural networks trained and validated on 64,800 labelled samples collected from instrumented corn and soybean field deployments over 180 days, achieving a classification accuracy of 91.3% and F1-score of 93.0% at an inference energy cost of 0.87 mJ per cycle. Circuit-level simulations of the 40 nm CMOS WuRX confirm an input return loss of -15.8 dB, transmission gain of +18.2 dB, and power-added efficiency of 45.3% at -60 dBm sensitivity, with continuous standby consumption of 45–53 µW. System-level power budget analysis validates battery lifetimes of 200–300 days using 240 mAh lithium cells augmented by a 5 cm² monocrystalline solar harvester. Key limitations include the restriction of field validation to two crop types and dependence on adequate solar irradiance for energy-neutral operation. The proposed architecture reduces communication overhead by 80–90% relative to threshold-based systems, providing a practical framework for scalable, resource-constrained precision agriculture deployments.","精准农业需要能够持续、能源自主地监测土壤，以检测诸如水分快速流失或养分耗竭接近威胁作物阈值等关键状况。传统无线传感器网络面临多年电池寿命与亚秒级响应延迟之间的根本性权衡，因为占空比轮询会在空闲查询上浪费能量，而始终在线的通信对于覆盖数百英亩的电池供电田间部署而言仍不可行。本研究提出了一种新颖的集成方案，将基于TinyML的边缘推理与定制设计的40 nm CMOS超低功耗唤醒接收器（WuRX）相结合，实现事件驱动的异步土壤状况监测，而无需依赖周期性轮询调度。所提出的三层边缘-雾-云架构采用测量水分、温度、pH、氮、磷和钾的传感器节点，其量化神经网络在180天内从仪器化玉米和大豆田间部署中采集的64,800个标记样本上进行训练和验证，在每次推理周期能耗为0.87 mJ的情况下，实现了91.3%的分类准确率和93.0%的F1分数。对40 nm CMOS WuRX的电路级仿真证实，在-60 dBm灵敏度下，输入回波损耗为-15.8 dB，传输增益为+18.2 dB，功率附加效率为45.3%，连续待机功耗为45–53 µW。系统级功率预算分析验证了使用240 mAh锂电池并辅以5 cm²单晶硅太阳能收集器时，电池寿命可达200–300天。主要局限性包括田间验证仅限于两种作物类型，以及能量中性运行依赖充足的太阳辐照度。与基于阈值的系统相比，所提出的架构将通信开销降低了80–90%，为可扩展、资源受限的精准农业部署提供了实用框架。","Informatyka Automatyka Pomiary w Gospodarce i Ochronie Środowiska","2026-09-30T00:00:00Z",80,{"impact":19,"substance":18,"depth":255,"authority":127,"freshness":256,"relevant":21,"comment":257},19,8,"将TinyML边缘推理与40nm CMOS超低功耗唤醒接收机结合，实现事件驱动型土壤监测，实测数据扎实、能耗指标明确，对精准农业物联网部署有参考价值。",[259],{"name":251,"url":248},[26,27,136,261,28],"无线传感器网络",{"code":263,"slug":264,"name":265,"path":266,"parent_code":117},"I5","agricultural-iot","物联网传感与边缘计算","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-iot",[268,269],{"code":38,"slug":42,"name":108,"path":109,"parent_code":110},{"code":270,"slug":271,"name":272,"path":273,"parent_code":110},"H8","sustainable-agriculture","可持续与保护性农业","\u002Ftopic\u002Fenvironment\u002Fsustainable-agriculture",{"status":119,"locked":14,"version":33,"method":34,"evidence":275},{"candidates":276,"agricultural_evidence":40},[277,284,288,293,297,304,311,318],{"code":263,"score":278,"supported":40,"title":279,"abstract":281,"tags":283},15,[280],"wireless sensor",[282,280],"无线传感",[261],{"code":38,"score":39,"supported":40,"title":285,"abstract":286,"tags":287},[42],[44,42],[],{"code":270,"score":256,"supported":40,"title":289,"abstract":291,"tags":292},[290],"sustainable agriculture",[],[],{"code":54,"score":256,"supported":40,"title":294,"abstract":295,"tags":296},[],[136,137],[26,136],{"code":298,"score":89,"supported":40,"title":299,"abstract":300,"tags":303},"A3",[],[301,302],"玉米","corn",[],{"code":305,"score":89,"supported":40,"title":306,"abstract":307,"tags":310},"A4",[],[308,309],"大豆","soybean",[],{"code":312,"score":313,"supported":40,"title":314,"abstract":315,"tags":317},"F5",3,[],[316],"harvester",[],{"code":47,"score":21,"supported":14,"title":319,"abstract":320,"tags":321},[],[],[27],[323,324],"TinyML 土壤监测","超低功耗 唤醒接收机 精准农业","TinyML土壤监测-4073","10.35784\u002Fiapgos.9467",{"doi":326,"openalex_id":328,"authors":329,"venue":251,"cited_by_count":78,"oa_url":333,"card":334,"direction":187,"ingested_from":86},"W7214927134",[330],{"name":331,"orcid":332},"Muhammad Ovais Akhter","https:\u002F\u002Forcid.org\u002F0000-0002-8928-4489","https:\u002F\u002Fph.pollub.pl\u002Findex.php\u002Fiapgos\u002Farticle\u002Fdownload\u002F9467\u002F5631",{"tldr":335,"method":336,"finding":337,"direction":187,"opportunity":338},"提出TinyML边缘推理与超低功耗唤醒接收机集成的土壤监测系统，实现事件驱动异步监测。","TinyML量化神经网络、40nm CMOS唤醒接收机、三层边缘-雾-云架构，6","分类准确率91.3%，通信开销降低80-90%，电池寿命200-300天。","可探索多作物多土壤类型泛化、无光照或弱光下能量收集与事件触发协同优化。","2026-10-03T23:30:17.793958Z",{"id":341,"title":342,"url":343,"summary":344,"summary_zh":345,"content":9,"source_name":346,"source_url":343,"published_at":347,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":348,"score_detail":349,"sources":353,"tags":355,"primary_topic":358,"secondary_topics":363,"classification":366,"search_phrases":392,"slug":394,"view_count":395,"doi":396,"paper":397,"created_at":413},2140,"AIOT in Predictive Agriculture - IoT and AI Integration for Real-Time Soil Monitoring and Smart Irrigation in Predictive Agriculture","https:\u002F\u002Fdoi.org\u002F10.22214\u002Fijraset.2026.84727","Modern agriculture faces severe challenges due to climate volatility, accelerating groundwater depletion, and the global imperative to maximize crop production on diminishing arable land. Traditional irrigation frameworks rely predominantly on static schedules or reactive threshold switching, leading to substantial water waste, energy inefficiencies, and suboptimal crop yields. To overcome these limitations, this paper proposes an end-to-end Artificial Intelligence of Things (AIoT) framework designed for real-time multi-parameter soil tracking and predictive smart irrigation. The system architecture deploys low-power IoT field nodes driven by ESP32 microcontrollers, integrated with capacitive soil moisture sensors, environmental sensors, and soil pH probes that stream telemetry data over lightweight MQTT protocols. To transition from reactive monitoring to proactive resource allocation, a cloud-based predictive engine utilizes Long Short-Term Memory (LSTM) neural networks to forecast 24- to-48-hour soil moisture depletion dynamics based on historical moisture profiles and localized meteorological factors. Experimental validation across a 90-day testbed demonstrates that the proposed predictive framework achieves a to reduction in total water consumption while maintaining optimal volumetric soil water content. Furthermore, deep-sleep dynamic power profiling confirms node energy autonomy of up to 219 days on a single battery charge, presenting a scalable, sustainable, and economically viable solution for precision agriculture.","现代农业正面临气候波动、地下水加速枯竭以及全球在日益减少的耕地上最大化作物产量的迫切需求等严峻挑战。传统灌溉框架主要依赖静态调度或反应式阈值切换，导致大量水资源浪费、能源效率低下以及作物产量欠优。为克服这些局限，本文提出了一种端到端的人工智能物联网（AIoT）框架，专为实时多参数土壤监测与预测性智能灌溉而设计。该系统架构部署了由ESP32微控制器驱动的低功耗物联网田间节点，集成了电容式土壤水分传感器、环境传感器和土壤pH探头，通过轻量级MQTT协议传输遥测数据。为实现从反应式监测向主动式资源分配的转变，基于云的预测引擎利用长短期记忆（LSTM）神经网络，根据历史水分剖面和局部气象因素，预测24至48小时的土壤水分消耗动态。在为期90天的测试平台上进行的实验验证表明，所提出的预测框架在保持最优土壤体积含水量的同时，实现了总用水量的降低。此外，深度睡眠动态功耗分析证实，节点在单次电池充电下可实现长达219天的能量自主运行，为精准农业提供了一种可扩展、可持续且经济可行的解决方案。","International Journal for Research in Applied Science and Engineering Technology","2026-09-10T00:00:00Z",75,{"impact":19,"substance":350,"depth":351,"authority":17,"freshness":256,"relevant":21,"comment":352},20,17,"AIoT+LSTM 预测灌溉的完整实证研究，90 天试验与节水、能耗数据扎实，对精准农业落地有参考价值，但期刊层级与影响范围偏细分领域。",[354],{"name":346,"url":343},[26,27,356,357,136,28],"物联网","智能灌溉",{"code":359,"slug":360,"name":361,"path":362,"parent_code":110},"H3","irrigation","水资源与灌溉","\u002Ftopic\u002Fenvironment\u002Firrigation",[364,365],{"code":263,"slug":264,"name":265,"path":266,"parent_code":117},{"code":38,"slug":42,"name":108,"path":109,"parent_code":110},{"status":119,"locked":14,"version":33,"method":34,"evidence":367},{"candidates":368,"agricultural_evidence":40},[369,375,380,384,388],{"code":359,"score":19,"supported":40,"title":370,"abstract":371,"tags":374},[360],[372,373,360],"灌溉","水资源",[357],{"code":263,"score":278,"supported":40,"title":376,"abstract":378,"tags":379},[377],"IoT",[356,377],[356],{"code":38,"score":39,"supported":40,"title":381,"abstract":382,"tags":383},[42],[44,42],[],{"code":54,"score":256,"supported":40,"title":385,"abstract":386,"tags":387},[],[136,137],[26,136],{"code":47,"score":20,"supported":40,"title":389,"abstract":390,"tags":391},[],[214],[27],[393,61],"农业人工智能 土壤监测 智慧农业 智能灌溉","农业人工智能土壤监测智慧农业智能灌溉-2140",5,"10.22214\u002Fijraset.2026.84727",{"doi":396,"openalex_id":398,"authors":399,"venue":346,"cited_by_count":78,"oa_url":343,"card":408,"direction":187,"ingested_from":86},"W7212115545",[400,402,404,406],{"name":401,"orcid":9},"Gowri M.",{"name":403,"orcid":9},"Boomika M.",{"name":405,"orcid":9},"S. Rakshana",{"name":407,"orcid":9},"Rubali R.",{"tldr":409,"method":410,"finding":411,"direction":187,"opportunity":412},"提出AIoT框架，用LSTM预测土壤湿度实现智能灌溉，节水并延长节点续航。","ESP32节点+电容湿度\u002FpH传感器，MQTT上云，LSTM预测24-48小时湿","90天试验节水显著，土壤含水量保持最优，单次电池续航达219天。","可探索多作物多气候下LSTM泛化能力，及边缘端轻量预测模型降低云依赖。","2026-09-11T23:30:10.173013Z",{"id":415,"title":416,"url":417,"summary":418,"summary_zh":419,"content":9,"source_name":420,"source_url":417,"published_at":421,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":422,"score_detail":423,"sources":425,"tags":427,"primary_topic":428,"secondary_topics":429,"classification":431,"search_phrases":457,"slug":459,"view_count":460,"doi":461,"paper":462,"created_at":474},1328,"Explainable Machine Learning for Simultaneous Prediction of Soil Texture and Gravimetric Water Content: A SHAP-Driven Multi-Model Framework","https:\u002F\u002Fdoi.org\u002F10.70917\u002Fijcisim-2026-5335","Soil texture and gravimetric water content (GWC) are important properties which need to be determined accurately for proper irrigation and sustainable land management. Although machine learning (ML) has enhanced the ability to predict pedometrics, many of the high performing algorithms have been described as a \"black-box\" and practitioners have been reluctant to trust the model and connect the model output with soil physics. Furthermore, the existing approaches mostly assume independence between the attributes to be predicted and do not take into account their physical relationships. The present study aims to fill these voids by proposing a robust multi-model approach for simultaneous prediction of soil texture and GWC. A variety of data was used to assess state-of-the-art algorithms, such as XGBoost, CatBoost, and Random Forest. However, to achieve local and global interpretability, we integrated Shapley Additive Explanations (SHAP). This XAI-based approach filled the gap between computational intelligence and soil physics, allowing a fine-grained diagnosis of the contribution of the different features. The results show that the Random Forest framework outperformed in terms of concurrent performance with R2 value of 0.908, accuracy of 90.8% and Cohen's Kappa of 0.82 for soil texture and R2 value of 0.798 for GWC. Two features, Cation Exchange Capacity (CEC) and Organic Matter Percentage (OMP), were identified as the most influential using SHAP analysis, and they were found to exert the greatest influence on the moisture retention thresholds, with CEC being the most discriminative feature between the Clayey and Sandy texture classes, which aligns with the model's validity in light of the known pedology. This transparency and simultaneous monitoring of key soil properties makes it an appealing solution for precision agriculture and for resource management in the context of climate change.","土壤质地与重量含水量（GWC）是精准灌溉和可持续土地管理中需精确测定的重要属性。尽管机器学习（ML）增强了预测土壤计量学（pedometrics）的能力，但许多高性能算法被描述为“黑箱”，实践者往往不愿信任模型，也难以将模型输出与土壤物理学联系起来。此外，现有方法大多假设待预测属性之间相互独立，未考虑其物理关系。本研究旨在通过提出一种稳健的多模型方法，同时预测土壤质地和GWC，以填补这些空白。研究使用了多种数据评估先进算法，如XGBoost、CatBoost和随机森林（Random Forest）。然而，为实现局部和全局可解释性，我们整合了Shapley加法解释（SHAP）。这种基于可解释人工智能（XAI）的方法弥合了计算智能与土壤物理学之间的鸿沟，能够对不同特征的贡献进行细粒度诊断。结果表明，随机森林框架在并发性能上表现优异，土壤质地的R²值为0.908，准确率为90.8%，Cohen's Kappa系数为0.82，GWC的R²值为0.798。通过SHAP分析，阳离子交换量（CEC）和有机质百分比（OMP）被识别为最具影响力的两个特征，且它们对水分保持阈值的影响最大，其中CEC是区分黏土和砂土质地类别的最具判别力的特征，这与已知土壤学知识下模型的效度相吻合。这种对关键土壤属性的透明性和同步监测，使其成为精准农业及气候变化背景下资源管理的理想解决方案。","International Journal of Computer Information Systems and Industrial Management Applications","2026-08-30T00:00:00Z",73,{"impact":19,"substance":18,"depth":19,"authority":17,"freshness":313,"relevant":21,"comment":424},"研究提出可解释的机器学习框架，同时预测土壤质地与含水量，提升模型透明度，对精准农业有实质贡献。",[426],{"name":420,"url":417},[26,27,142,28],{"code":47,"slug":203,"name":204,"path":205,"parent_code":117},[430],{"code":38,"slug":42,"name":108,"path":109,"parent_code":110},{"status":119,"locked":14,"version":33,"method":34,"evidence":432},{"candidates":433,"agricultural_evidence":40},[434,438,442,446,450],{"code":47,"score":212,"supported":40,"title":435,"abstract":436,"tags":437},[143],[142,143],[27,142],{"code":38,"score":39,"supported":40,"title":439,"abstract":440,"tags":441},[42],[44,42],[],{"code":54,"score":133,"supported":40,"title":443,"abstract":444,"tags":445},[],[136,137],[26],{"code":359,"score":89,"supported":40,"title":447,"abstract":448,"tags":449},[],[372,360],[],{"code":451,"score":89,"supported":40,"title":452,"abstract":453,"tags":456},"H7",[],[454,455],"气候变化","climate change",[],[458,61],"农业人工智能 土壤监测 智慧农业 机器学习","农业人工智能土壤监测智慧农业机器学习-1328",2,"10.70917\u002Fijcisim-2026-5335",{"doi":461,"openalex_id":463,"authors":464,"venue":420,"cited_by_count":78,"oa_url":417,"card":469,"direction":241,"ingested_from":86},"W7204858035",[465,467],{"name":466,"orcid":9},"Chetana Shivanagi",{"name":468,"orcid":9},"Shivayogi  Ullagaddi",{"tldr":470,"method":471,"finding":472,"direction":187,"opportunity":473},"提出可解释机器学习框架，同时预测土壤质地和含水量，提升模型透明度和可信度。","使用XGBoost、CatBoost、随机森林，结合SHAP进行局部和全局解释。","随机森林表现最佳，CEC和有机质是关键特征，与土壤物理规律一致。","可探索将SHAP解释与因果推断结合，或扩展到其他土壤属性及动态预测，增强模型泛化性。","2026-09-01T23:30:40.247344Z",{"id":476,"title":477,"url":478,"summary":479,"summary_zh":480,"content":9,"source_name":481,"source_url":478,"published_at":482,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":483,"score_detail":484,"sources":487,"tags":489,"primary_topic":493,"secondary_topics":499,"classification":501,"search_phrases":518,"slug":521,"view_count":78,"doi":522,"paper":523,"created_at":537},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":256,"substance":485,"depth":19,"authority":17,"freshness":13,"relevant":21,"comment":486},21,"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。",[488],{"name":481,"url":478},[26,27,490,491,492],"小农户","植物病害识别","可解释性",{"code":494,"slug":495,"name":496,"path":497,"parent_code":498},"G2","farm-services","社会化服务与规模经营","\u002Ftopic\u002Frural-services\u002Ffarm-services","G",[500],{"code":47,"slug":203,"name":204,"path":205,"parent_code":117},{"status":119,"locked":14,"version":33,"method":34,"evidence":502},{"candidates":503,"agricultural_evidence":40},[504,509,514],{"code":494,"score":17,"supported":40,"title":505,"abstract":507,"tags":508},[506],"smallholder",[506],[490],{"code":47,"score":13,"supported":40,"title":510,"abstract":512,"tags":513},[511],"computer vision",[],[27,491],{"code":54,"score":21,"supported":14,"title":515,"abstract":516,"tags":517},[],[],[26],[519,520],"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":522,"openalex_id":524,"authors":525,"venue":481,"cited_by_count":78,"oa_url":478,"card":532,"direction":241,"ingested_from":86},"W7223137323",[526,528,530],{"name":527,"orcid":9},"Akshay Ramgoolam",{"name":529,"orcid":9},"Priya Bundhoo Appadoo",{"name":531,"orcid":9},"Jean-Noël LEspérance",{"tldr":533,"method":534,"finding":535,"direction":241,"opportunity":536},"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"]