[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-3009":3,"related-3009":55},{"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":54},3009,"AgriMAC: An Attention Based Multimodal Deep Clustering Framework for Rice Health Assessment","https:\u002F\u002Fdoi.org\u002F10.22266\u002Fijies2026.1031.06","Rice is Indonesia's staple crop, yet its productivity has declined in recent years because pest and disease outbreaks remain difficult to detect at an early stage.Existing precision agriculture approaches commonly process Internet of Things (IoT) sensor data and remote sensing imagery independently and often rely on supervised learning, requiring large amounts of labeled data.Meanwhile, multispectral drone imagery producing the Normalized Difference Vegetation Index (NDVI) provides richer information on crop physiological conditions than RGB-based vegetation indices.This study proposes Agricultural Multimodal Attention Clustering (AgriMAC), an unsupervised framework that integrates UAV derived NDVI imagery, 7-in-1 IoT soil sensor measurements, and historical weather data from the Open-Meteo API for rice field condition monitoring.Each modality is encoded using a dedicated autoencoder and fused through an entropy-regularized attention mechanism before Deep Embedded Clustering is performed.To reduce the influence of crop growth stage, the IoT representation is residualized using growth-phase statistics estimated exclusively from the training fold, enabling the discovered clusters to represent within-phase agronomic conditions rather than crop age.Experiments conducted under a grouped leave-one-field-out protocol produced a Silhouette Score of 0.465 ± 0.048, a Davies Bouldin Index of 0.807 ± 0.036, and a Calinski Harabasz Index of 195 ± 27.The learned groups also showed low normalized mutual information with growth phase (0.079 ± 0.043) and near chance phase decodability (balanced accuracy = 0.554 ± 0.042), indicating that they are only weakly associated with crop growth stage.The learned attention weights identified IoT soil measurements (0.570 ± 0.024) as the dominant modality, while NDVI imagery (0.210 ± 0.014) and weather information (0.220 ± 0.014) provided complementary spatial and temporal context.Overall, AgriMAC provides an interpretable and leakage-aware framework for multimodal clustering of rice field conditions.Although its clustering performance is comparable to that of a capacity-matched IoT-only model, it additionally quantifies the contribution of each sensing modality through attention weights and explicitly mitigates the growth-phase confound, making it suitable for field level agronomic condition monitoring and spatial decision support in precision agriculture.","水稻是印度尼西亚的主要作物，但近年来其生产力有所下降，因为病虫害暴发在早期阶段仍难以检测。现有的精准农业方法通常独立处理物联网（IoT）传感器数据和遥感影像，且往往依赖监督学习，需要大量标注数据。与此同时，生成归一化植被指数（NDVI）的多光谱无人机影像比基于RGB的植被指数能提供更丰富的作物生理状况信息。本研究提出农业多模态注意力聚类（AgriMAC），这是一个无监督框架，整合了无人机获取的NDVI影像、七合一IoT土壤传感器测量数据以及来自Open-Meteo API的历史天气数据，用于稻田状况监测。每种模态均使用专用自编码器进行编码，并通过熵正则化注意力机制进行融合，随后执行深度嵌入聚类。为减少作物生长阶段的影响，IoT表征利用仅从训练折估计的生长阶段统计量进行残差化处理，使发现的聚类能够表征阶段内的农艺状况而非作物年龄。在分组留一田块协议下进行的实验产生了0.465 ± 0.048的轮廓系数、0.807 ± 0.036的Davies-Bouldin指数和195 ± 27的Calinski-Harabasz指数。学习到的分组还显示出与生长阶段的低归一化互信息（0.079 ± 0.043）以及接近随机的阶段可解码性（平衡准确率 = 0.554 ± 0.042），表明它们与作物生长阶段仅存在弱关联。学习到的注意力权重将IoT土壤测量（0.570 ± 0.024）识别为主导模态，而NDVI影像（0.210 ± 0.014）和天气信息（0.220 ± 0.014）则提供了互补的空间和时间背景。总体而言，AgriMAC为稻田状况的多模态聚类提供了一个可解释且感知数据泄漏的框架。尽管其聚类性能与容量匹配的仅IoT模型相当，但它还通过注意力权重量化了每种传感模态的贡献，并明确减轻了生长阶段混杂因素，使其适用于田块级农艺状况监测和精准农业中的空间决策支持。",null,"International journal of intelligent engineering and systems","2026-09-19T00:00:00Z","论文",10,false,72,{"impact":17,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":23},12,21,17,13,9,1,"提出无监督多模态注意力聚类框架，融合无人机NDVI、IoT土壤与气象数据评估水稻健康，方法新颖且实验严谨，对精准农业田间监测有参考价值。",[25],{"name":10,"url":6},[27,28,29,30,31],"智慧农业","农业人工智能","水稻","多模态融合","遥感",[33,34],"AgriMAC 水稻 多模态聚类","无人机 NDVI 水稻 病害监测","AgriMAC水稻多模态聚类-3009",0,"10.22266\u002Fijies2026.1031.06",{"doi":37,"openalex_id":39,"authors":40,"venue":10,"cited_by_count":36,"oa_url":6,"card":47,"direction":51,"ingested_from":53},"W7213619014",[41,43,45],{"name":42,"orcid":9},"Nurfadhilah Mardianti Andini",{"name":44,"orcid":9},"Mike Yuliana",{"name":46,"orcid":9},"Moch. Zen Samsono Hadi",{"tldr":48,"method":49,"finding":50,"direction":51,"opportunity":52},"提出无监督多模态聚类框架AgriMAC，融合无人机NDVI、IoT土壤与气象数据评估水稻健康。","各模态自编码器编码，熵正则注意力融合，深度嵌入聚类，按生长阶段残差化。","聚类性能与仅IoT模型相当，但注意力权重可解释模态贡献并弱化生长阶段混淆。","智慧农业 \u002F 农业物联网","可探索注意力融合机制在更多作物与传感器组合下的泛化性，并引入时序动态聚类。","openalex","2026-09-20T23:30:08.419613Z",{"total":56,"page":22,"page_size":56,"items":57},6,[58,115,162,202,251,305],{"id":59,"title":60,"url":61,"summary":62,"summary_zh":63,"content":9,"source_name":64,"source_url":61,"published_at":65,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":66,"score_detail":67,"sources":70,"tags":72,"search_phrases":74,"slug":77,"view_count":36,"doi":78,"paper":79,"created_at":114},2512,"Deep learning-based classification of wet direct seeded rice, broadcasted direct seeded rice and transplanted rice using drone imagery for precision agriculture","https:\u002F\u002Fdoi.org\u002F10.3389\u002Ffrsen.2026.1698781","Accurate estimation of crop area using classification techniques applied to drone imagery plays an important role in precision agriculture. Traditional machine learning (ML) approaches have been widely used for agricultural image classification; however, advanced deep learning (DL) models generally provide superior feature extraction and classification capability for complex crop patterns. Differentiating various rice establishment methods is essential for precise area estimation. In this study, advanced deep learning models were employed to classify three types of rice cultivation: (i) Broadcasted Direct Seeded Rice (DSR), (ii) Wet Direct Seeded Rice (Wet DSR), and (iii) Transplanted Rice (TR) using drone imagery. The drone imagery was collected from an experimental field at Praanadhaara Organised Agro Forestry Private Limited, Bapatla District, Andhra Pradesh, India. The images were captured in the visible spectrum (Red, Green, and Blue bands) on three dates, viz., 15 October 2023, 27 October 2023, and 01 December 2023, from an altitude of 40 m and were used to train and test classification models. Classification was performed using six ResNet-50 based hybrid models, namely, ResNet-50+K-Nearest Neighbor (ResNet-50+KNN), ResNet-50+Support Vector Machine (ResNet-50+SVM), ResNet-50+Decision Trees (ResNet-50+DT), ResNet-50+Random Forest (ResNet-50+RF), ResNet-50+Naïve Bayes (ResNet-50+NB), and ResNet-50+Neural Network (ResNet-50+NN), along with two additional DL architectures, namely, You Only Look Once version 5 (YOLOv5) and You Only Look Once version 8 (YOLOv8). Model performance was evaluated using overall accuracy (OA), precision (P), recall (R), kappa coefficient (K), F1-score (F1), and mean Average Precision (mAP). Among the tested classifiers, the ResNet-50+NN model consistently achieved the highest average overall accuracies of 93.16%, 95.81%, and 93.31% at T 1 , T 2 , and T 3 , respectively, outperforming all other models, whose accuracies ranged from 78.05% to 92.07%, 86.11%–95.24%, and 77.47%–92.09% across the respective time intervals. The ResNet-50+NN model also recorded the highest precision (0.93–0.97), recall (0.92–0.97), F1-score (0.92–0.97), and kappa coefficient (0.89–0.96), demonstrating superior and consistent classification performance across all observation dates. The methodology developed in this work enables precise identification of rice establishment methods, improving crop mapping and monitoring. This identification enhances resource efficiency, optimizes input use, and supports site-specific management, contributing to sustainable precision agriculture.","利用分类技术对无人机影像进行作物面积精确估算，在精准农业中发挥着重要作用。传统机器学习（ML）方法已广泛用于农业图像分类；然而，先进的深度学习（DL）模型通常对复杂作物模式具有更优越的特征提取和分类能力。区分不同的水稻种植方式对于精确估算面积至关重要。本研究采用先进的深度学习模型，利用无人机影像对三种水稻种植类型进行分类：（i）撒播直播稻（DSR），（ii）湿润直播稻（Wet DSR），以及（iii）移栽稻（TR）。无人机影像采集自印度安得拉邦巴帕特拉县Praanadhaara Organised Agro Forestry Private Limited的试验田。图像在可见光谱（红、绿、蓝波段）下于三个日期拍摄，即2023年10月15日、2023年10月27日和2023年12月1日，飞行高度为40 m，用于训练和测试分类模型。分类采用六种基于ResNet-50的混合模型，即ResNet-50+K近邻（ResNet-50+KNN）、ResNet-50+支持向量机（ResNet-50+SVM）、ResNet-50+决策树（ResNet-50+DT）、ResNet-50+随机森林（ResNet-50+RF）、ResNet-50+朴素贝叶斯（ResNet-50+NB）和ResNet-50+神经网络（ResNet-50+NN），以及两种额外的深度学习架构，即You Only Look Once第5版（YOLOv5）和You Only Look Once第8版（YOLOv8）。采用总体精度（OA）、精确率（P）、召回率（R）、Kappa系数（K）、F1分数（F1）和平均精度均值（mAP）评估模型性能。在测试的分类器中，ResNet-50+NN模型在T₁、T₂和T₃分别持续取得最高的平均总体精度，为93.16%、95.81%和93.31%，优于所有其他模型，后者的精度在相应时间段分别为78.05%–92.07%、86.11%–95.24%和77.47%–92.09%。ResNet-50+NN模型还记录了最高的精确率（0.93–0.97）、召回率（0.92–0.97）、F1分数（0.92–0.97）和Kappa系数（0.89–0.96），在所有观测日期均表现出优越且稳定的分类性能。本研究开发的方法能够精确识别水稻种植方式，改进作物制图和监测。这种识别增强了资源","Frontiers in Remote Sensing","2026-09-14T00:00:00Z",76,{"impact":68,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":69},16,"基于无人机可见光影像与ResNet-50混合模型区分三类水稻种植方式，方法对比系统、精度数据扎实，对精准农业作物制图有参考价值，但属细分技术进展，影响范围有限。",[71],{"name":64,"url":61},[27,28,29,73,31],"精准农业",[75,76],"农业人工智能 智慧农业 精准农业 水稻","农业人工智能 智慧农业","农业人工智能智慧农业精准农业水稻-2512","10.3389\u002Ffrsen.2026.1698781",{"doi":78,"openalex_id":80,"authors":81,"venue":64,"cited_by_count":36,"oa_url":107,"card":108,"direction":51,"ingested_from":53},"W7212834743",[82,84,86,88,90,93,95,97,99,102,104],{"name":83,"orcid":9},"Amrutha Lakshmi Gubbala",{"name":85,"orcid":9},"Santosha Rathod",{"name":87,"orcid":9},"Ramesh Dasyam",{"name":89,"orcid":9},"Mahender Kumar Rapolu",{"name":91,"orcid":92},"Arun Kumar Dasari","https:\u002F\u002Forcid.org\u002F0000-0001-7398-8601",{"name":94,"orcid":9},"Pundarikakshudu Kurra",{"name":96,"orcid":9},"Hanuma Raviteja Madireddy",{"name":98,"orcid":9},"Prajwal R. Shashishekhar",{"name":100,"orcid":101},"Ravi V. Mural","https:\u002F\u002Forcid.org\u002F0000-0002-5489-9918",{"name":103,"orcid":9},"Anil Kumar",{"name":105,"orcid":106},"R. M. Sundaram","https:\u002F\u002Forcid.org\u002F0000-0002-9857-8251","https:\u002F\u002Fwww.frontiersin.org\u002Fjournals\u002Fremote-sensing\u002Farticles\u002F10.3389\u002Ffrsen.2026.1698781\u002Fpdf",{"tldr":109,"method":110,"finding":111,"direction":112,"opportunity":113},"用无人机可见光影像和深度学习模型区分直播稻、湿直播稻与移栽稻三种水稻种植方式。","无人机RGB影像三个时相，比较ResNet-50混合模型与YOLOv5\u002Fv8分类","ResNet-50+NN表现最优，三期总体精度达93.16%、95.81%、93.31%，优于其他模","农业遥感与作物表型","可扩展到多光谱\u002F多时相与更大区域验证，并用于种植方式制图与面积精准估算。","2026-09-15T23:30:08.600535Z",{"id":116,"title":117,"url":118,"summary":119,"summary_zh":9,"content":9,"source_name":120,"source_url":118,"published_at":121,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":122,"score_detail":123,"sources":128,"tags":130,"search_phrases":132,"slug":134,"view_count":36,"doi":135,"paper":136,"created_at":161},2287,"UAV-based rice nitrogen status monitoring using machine learning and deep learning","https:\u002F\u002Fdoi.org\u002F10.1007\u002Fs11119-026-10440-8","UAV-based rice nitrogen status monitoring using machine learning and deep learning。Precision Agriculture","Precision Agriculture","2026-09-11T00:00:00Z",75,{"impact":68,"substance":124,"depth":19,"authority":125,"freshness":126,"relevant":22,"comment":127},20,14,8,"核心期刊论文，将机器学习与深度学习用于无人机水稻氮素诊断，方法新颖、结论可靠，对精准施肥有参考价值，但属细分技术进展，未达重大突破层级。",[129],{"name":120,"url":118},[27,28,29,31,131],"氮素监测",[133,76],"农业人工智能 智慧农业 氮素监测 水稻","农业人工智能智慧农业氮素监测水稻-2287","10.1007\u002Fs11119-026-10440-8",{"doi":135,"openalex_id":137,"authors":138,"venue":120,"cited_by_count":36,"oa_url":9,"card":9,"direction":112,"ingested_from":53},"W7212268243",[139,141,144,146,148,151,153,155,158],{"name":140,"orcid":9},"Xi Tao",{"name":142,"orcid":143},"Jiayi Zhang","https:\u002F\u002Forcid.org\u002F0000-0002-0401-996X",{"name":145,"orcid":9},"Zhaopeng Fu",{"name":147,"orcid":9},"Xinye Xu",{"name":149,"orcid":150},"Qiang Cao","https:\u002F\u002Forcid.org\u002F0000-0003-3733-2968",{"name":152,"orcid":9},"Yongchao Tian",{"name":154,"orcid":9},"Yan Zhu",{"name":156,"orcid":157},"Weixing Cao","https:\u002F\u002Forcid.org\u002F0000-0003-2622-7986",{"name":159,"orcid":160},"Xiaojun Liu","https:\u002F\u002Forcid.org\u002F0000-0001-7593-085X","2026-09-13T23:30:03.209118Z",{"id":163,"title":164,"url":165,"summary":166,"summary_zh":167,"content":9,"source_name":168,"source_url":165,"published_at":169,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":66,"score_detail":170,"sources":173,"tags":175,"search_phrases":177,"slug":180,"view_count":36,"doi":181,"paper":182,"created_at":201},2160,"An Optimized Temporal Attention-based Deep Learning Framwork for Accurate Crop Yield Prediction from Multi-source Agricultural Data","https:\u002F\u002Fdoi.org\u002F10.18805\u002Fag.d-6558","Background: In precision agriculture, crop yield prediction is crucial for supporting effective management of crops, timely settlement of farmers crop insurances, food security planning and serve as decision tool for sustainable agricultural achievements. However, prediction of crop yield is difficult due to the complex interaction behavior among diverse conditions in weather, soil properties, multispectral remote sensing observations and the dynamics in crop growth. Despite existing AI techniques like machine learning and deep learning models outperformed in predicting crop yield, they are often not able to capture the temporal dependencies. Furthermore, determining optimal hyperparameters for model training remains a challenging task, significantly influencing prediction accuracy, model convergence and generalization performance. Methods: To overcome this, the present study aimed to predict the crop yield using temporal attention (TA) based Artificial bee colony (ABC) optimized Bi - LSTM model using multi source agricultural data. The proposed model learns the sequential and temporal dependencies from both past and future time steps simultaneously, from the historical weather variables, soil properties, remote sensing derived normalized difference vegetation indices (NDVI) and historical yield data. The ABC algorithm was employed to automatically optimize the model hyperparameters with minimum run time. Result: The model is trained using district wise rice yield records of Tamil Nadu, historical weather variables, Sentinel-2 derived NDVI and soil properties. The dataset collected for major rice growing districts from 1995 to 2025 (30 years). The model shows the superior performance over conventional methods like random forest, support vector regression, deep neural network, convolutional neural network, long short-term memory models with lesser root mean square error (RMSE = 298.01 kg ha-1) and higher coefficient of determination, R2 (81.76%). Also, the training time of TA - ABC - Bi-LSTM model is comparatively less than the comparative models because of ABC optimization algorithm tuned the best hyperparameters compared to traditional trial and error manual method.","背景：在精准农业中，作物产量预测对于支持有效的作物管理、农民作物保险的及时理赔、粮食安全规划以及作为实现可持续农业的决策工具至关重要。然而，由于天气、土壤属性、多光谱遥感观测等多种条件之间复杂的交互作用以及作物生长的动态变化，作物产量预测仍然困难。尽管现有的机器学习、深度学习等人工智能技术在作物产量预测方面表现优异，但往往无法捕捉时间依赖性。此外，确定模型训练的最优超参数仍是一项具有挑战性的任务，显著影响预测精度、模型收敛性和泛化性能。方法：为解决上述问题，本研究旨在利用基于时间注意力（TA）的人工蜂群（ABC）算法优化的双向长短期记忆（Bi-LSTM）模型，结合多源农业数据预测作物产量。所提出的模型能够同时从历史和未来时间步中学习序列和时间依赖关系，数据来源包括历史气象变量、土壤属性、遥感衍生的归一化植被指数（NDVI）以及历史产量数据。采用ABC算法以最短运行时间自动优化模型超参数。结果：该模型使用泰米尔纳德邦各地区的稻米产量记录、历史气象变量、Sentinel-2衍生的NDVI和土壤属性进行训练。数据集收集了1995年至2025年（30年）主要水稻种植区的数据。该模型相比随机森林、支持向量回归、深度神经网络、卷积神经网络、长短期记忆模型等传统方法表现出更优的性能，均方根误差更低（RMSE = 298.01 kg ha-1），决定系数R²更高（81.76%）。此外，TA-ABC-Bi-LSTM模型的训练时间相比对比模型更短，因为ABC优化算法相比传统的试错手动方法能够调优出最佳超参数。","Agricultural Science Digest - A Research Journal","2026-09-10T00:00:00Z",{"impact":68,"substance":18,"depth":171,"authority":17,"freshness":21,"relevant":22,"comment":172},18,"提出时序注意力结合人工蜂群优化的Bi-LSTM产量预测模型，基于30年多源数据取得较高精度，方法新颖且数据规模可观，对精准农业与农业保险决策有参考价值。",[174],{"name":168,"url":165},[27,28,29,31,176],"作物产量预测",[178,179],"作物产量预测 农业人工智能 智慧农业 水稻","作物产量预测 农业人工智能","作物产量预测农业人工智能智慧农业水稻-2160","10.18805\u002Fag.d-6558",{"doi":181,"openalex_id":183,"authors":184,"venue":168,"cited_by_count":36,"oa_url":165,"card":195,"direction":112,"ingested_from":53},"W7212119230",[185,187,189,191,193],{"name":186,"orcid":9},"M. Lokeshwari",{"name":188,"orcid":9},"Nivethitha Manavalagan",{"name":190,"orcid":9},"P. Amrutha",{"name":192,"orcid":9},"S. Harinee",{"name":194,"orcid":9},"N.R. Divyasree",{"tldr":196,"method":197,"finding":198,"direction":199,"opportunity":200},"提出TA-ABC-Bi-LSTM模型，用多源农业数据精准预测水稻产量。","基于时间注意力与人工蜂群优化的Bi-LSTM，融合气象、土壤、Sentinel-","模型RMSE为298.01 kg\u002Fha，R²达81.76%，优于RF、SVR、DNN、CNN、LST","农业人工智能与决策模型","可探索轻量化注意力机制与迁移学习，将模型推广至多作物、小样本及实时预测场景。","2026-09-11T23:30:29.551995Z",{"id":203,"title":204,"url":205,"summary":206,"summary_zh":9,"content":9,"source_name":207,"source_url":205,"published_at":208,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":209,"score_detail":210,"sources":214,"tags":216,"search_phrases":217,"slug":219,"view_count":36,"doi":220,"paper":221,"created_at":250},1196,"Rice seedling diagnosis of light and water stress based on green – blue channel feature enhancement","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.compag.2026.112359","Rice seedling diagnosis of light and water stress based on green – blue channel feature enhancement。Computers and Electronics in Agriculture","Computers and Electronics in Agriculture","2026-08-29T00:00:00Z",60,{"impact":17,"substance":211,"depth":68,"authority":125,"freshness":212,"relevant":22,"comment":213},15,3,"基于绿蓝通道特征增强的水稻光水胁迫诊断方法，发表于核心期刊，具专业深度，但影响范围有限。",[215],{"name":207,"url":205},[27,28,29,31],[218,76],"农业人工智能 智慧农业 水稻 遥感","农业人工智能智慧农业水稻遥感-1196","10.1016\u002Fj.compag.2026.112359",{"doi":220,"openalex_id":222,"authors":223,"venue":207,"cited_by_count":36,"oa_url":244,"card":245,"direction":199,"ingested_from":53},"W4409570098",[224,227,230,233,235,238,241],{"name":225,"orcid":226},"Shaowen Liu","https:\u002F\u002Forcid.org\u002F0000-0002-9358-9648",{"name":228,"orcid":229},"Liwei Wang","https:\u002F\u002Forcid.org\u002F0000-0001-9970-8604",{"name":231,"orcid":232},"Zhigang Zhang","https:\u002F\u002Forcid.org\u002F0000-0002-8232-4900",{"name":234,"orcid":9},"HaoRan Xu",{"name":236,"orcid":237},"Yubo Yang","https:\u002F\u002Forcid.org\u002F0000-0002-9545-9345",{"name":239,"orcid":240},"Rui Gao","https:\u002F\u002Forcid.org\u002F0000-0002-4351-6555",{"name":242,"orcid":243},"Zhongbin Su","https:\u002F\u002Forcid.org\u002F0000-0002-8966-8933","https:\u002F\u002Fdoi.org\u002F10.2139\u002Fssrn.5222086",{"tldr":246,"method":247,"finding":248,"direction":199,"opportunity":249},"提出基于绿蓝通道特征增强的水稻幼苗光水胁迫诊断方法。","利用绿蓝通道特征增强技术分析图像，诊断胁迫。","绿蓝通道特征增强可有效识别水稻幼苗光水胁迫。","可探索多胁迫联合诊断及田间实时监测系统，结合深度学习提升精度。","2026-09-01T04:03:01.818562Z",{"id":252,"title":253,"url":254,"summary":255,"summary_zh":256,"content":9,"source_name":207,"source_url":254,"published_at":11,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":257,"score_detail":258,"sources":261,"tags":263,"search_phrases":266,"slug":269,"view_count":36,"doi":270,"paper":271,"created_at":304},3006,"Explainable spectral–image fusion multi-task learning for maize canopy biochemical and structural trait retrieval","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.compag.2026.112450","Accurate and interpretable estimation of multiple maize canopy traits is important for proximal crop monitoring and high-throughput phenotyping. This study developed a spectral–image fusion multi-task learning framework (MM-MTL) for the simultaneous retrieval of chlorophyll index (Chl index), leaf area index (LAI), nitrogen balance index (NBI), and anthocyanin index (Anth index). Proximal hyperspectral observations were collected using a Specim IQ camera covering 400–1000 nm with 204 spectral bands. For each observation, an ROI-mean spectral vector was extracted to retain fine-grained canopy reflectance information, while a co-registered pseudo-RGB image derived from visible bands of the same hyperspectral cube preserved two-dimensional canopy structure and visible appearance. MM-MTL integrates spectral and image feature extraction, task-specific fusion, and uncertainty-weighted multi-task learning to jointly estimate the four traits. A total of 1,023 valid samples collected from nine field campaigns across the 2024 and 2025 growing seasons were used for model development and evaluation. Under plot-grouped five-fold cross-validation, MM-MTL achieved R 2 values of 0.872, 0.885, 0.722, and 0.807 for Chl index, LAI, NBI, and Anth index, respectively, and consistently outperformed the single-task, single-representation, and conventional regression baselines. Performance decreased under more challenging generalization settings, with R 2 values ranging from 0.543 to 0.680 under leave-one-campaign-out validation and from 0.425 to 0.640 under bidirectional cross-year validation. Ablation and post-hoc analyses further showed that preserving image spatial organization improved prediction, while task-wise routing, input-representation masking, wavelength perturbation, and cross-fold stability analyses revealed trait-dependent use of spectral and spatial information. These results demonstrate that complementary spectral and spatial representations derived from the same hyperspectral observation can improve multi-trait maize canopy retrieval, while transfer across acquisition campaigns, years, and field environments remains an important direction for further improvement.","准确且可解释地估算多个玉米冠层性状，对于近地作物监测和高通量表型分析具有重要意义。本研究开发了一种光谱—图像融合多任务学习框架（MM-MTL），用于同时反演叶绿素指数（Chl index）、叶面积指数（LAI）、氮平衡指数（NBI）和花青素指数（Anth index）。近地高光谱观测使用Specim IQ相机采集，覆盖400–1000 nm，共204个光谱波段。对于每次观测，提取ROI均值光谱向量以保留细粒度冠层反射率信息，同时从同一高光谱立方体的可见光波段生成配准的伪RGB图像，以保留二维冠层结构和可见外观。MM-MTL集成了光谱与图像特征提取、任务特定融合以及不确定性加权多任务学习，以联合估算这四种性状。研究使用2024年和2025年生长季9次田间试验采集的共计1,023个有效样本进行模型开发与评估。在按小区分组的五折交叉验证下，MM-MTL对Chl index、LAI、NBI和Anth index的R²分别为0.872、0.885、0.722和0.807，且持续优于单任务、单表征和传统回归基线。在更具挑战性的泛化设置下，模型性能有所下降，留一试验验证的R²范围为0.543–0.680，双向跨年验证的R²范围为0.425–0.640。消融分析和事后分析进一步表明，保留图像空间组织可提升预测性能，而任务路由、输入表征掩蔽、波长扰动和跨折稳定性分析揭示了光谱与空间信息的性状依赖性利用方式。这些结果表明，从同一高光谱观测中提取的互补光谱与空间表征可改善多性状玉米冠层反演，而跨采集试验、年份和田间环境的迁移仍是未来改进的重要方向。",81,{"impact":171,"substance":259,"depth":171,"authority":125,"freshness":21,"relevant":22,"comment":260},22,"提出光谱-图像融合多任务学习框架，1023份样本跨两年验证，方法新颖且结论可靠，对作物高通量表型研究有实质参考价值。",[262],{"name":207,"url":254},[27,28,264,31,265],"玉米","高通量表型",[267,268],"玉米冠层 多任务学习 高光谱","Specim IQ 玉米 表型","玉米冠层多任务学习高光谱-3006","10.1016\u002Fj.compag.2026.112450",{"doi":270,"openalex_id":272,"authors":273,"venue":207,"cited_by_count":36,"oa_url":254,"card":299,"direction":112,"ingested_from":53},"W7213658705",[274,276,278,280,282,285,287,290,292,294,296],{"name":275,"orcid":9},"Penglei Zhang",{"name":277,"orcid":9},"Tianbo Hao",{"name":279,"orcid":9},"Zhuoyuan Zhao",{"name":281,"orcid":9},"Hong Sun",{"name":283,"orcid":284},"Yelu Zeng","https:\u002F\u002Forcid.org\u002F0000-0003-4267-1841",{"name":286,"orcid":9},"Zheng Cui",{"name":288,"orcid":289},"Ta Na","https:\u002F\u002Forcid.org\u002F0000-0002-1348-5655",{"name":291,"orcid":9},"Lang Qiao",{"name":293,"orcid":9},"Durval Dourado Neto",{"name":295,"orcid":9},"Feng Yang",{"name":297,"orcid":298},"Jingzhu Wu","https:\u002F\u002Forcid.org\u002F0000-0002-8386-1038",{"tldr":300,"method":301,"finding":302,"direction":112,"opportunity":303},"提出光谱-图像融合多任务学习框架，同时反演玉米冠层四个生化与结构性状。","Specim IQ高光谱与伪RGB融合，不确定性加权多任务学习，1023样本交叉","融合模型精度优于单任务基线，但跨年份与跨环境泛化性能明显下降。","可研究跨年份\u002F跨环境域适应与迁移学习，提升多性状反演泛化能力。","2026-09-20T23:30:01.739142Z",{"id":306,"title":307,"url":308,"summary":309,"summary_zh":9,"content":9,"source_name":310,"source_url":9,"published_at":311,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":312,"score_detail":313,"sources":315,"tags":317,"search_phrases":320,"slug":323,"view_count":36,"doi":9,"paper":324,"created_at":332},2999,"基于改进DeepLabv3+的高标准农田田间道路提取与结构指标量化框架","https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F18\u002F1986","沈阳农业大学刘永生等开发了基于MobileNetV2改进DeepLabv3+的高标准农田田间道路提取轻量化框架，集成Normalization-based Attention Module与Content-Aware ReAssembly of FEatures。三次随机种子训练平均mIoU 93.34%、mPA 96.75%、精度98.90%，模型参数6.14M、推理速度17.04 FPS；沥青、混凝土、砾石道路宽度预测R²分别为0.650、0.486、0.662，宽度MAE 0.130\u002F0.140\u002F0.100 m。第二验证区域连通性指数从0.4682提升至0.4795，支持高标准农田田间道路高效、可量化、可追溯的验收检查。","MDPI Agriculture 16(18):1986","2026-09-16T00:00:00Z",77,{"impact":68,"substance":259,"depth":171,"authority":20,"freshness":126,"relevant":22,"comment":314},"方法有创新、指标详实，对高标准农田道路验收有实用价值，但属细分技术论文，影响面有限。",[316],{"name":310,"url":308},[27,28,318,31,319],"高标准农田","田间道路",[321,322],"沈阳农业大学 高标准农田 道路提取","DeepLabv3 田间道路 遥感","沈阳农业大学高标准农田道路提取-2999",{"doi":9,"openalex_id":9,"authors":325,"venue":9,"cited_by_count":36,"oa_url":9,"card":326,"direction":112,"ingested_from":331},[],{"tldr":327,"method":328,"finding":329,"direction":112,"opportunity":330},"提出改进DeepLabv3+轻量框架，提取高标准农田田间道路并量化结构指标。","MobileNetV2+NAM+CARAFE改进DeepLabv3+，多区域遥感","mIoU 93.34%，道路宽度预测R²最高0.662，连通性指数提升至0.4795。","可拓展至多作物、多地形道路提取，并结合时序遥感实现道路损毁动态监测。","agent","2026-09-20T00:03:08.023498Z"]