[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-4822":3,"related-4822":93},{"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,"primary_topic":32,"secondary_topics":38,"classification":39,"search_phrases":67,"slug":70,"view_count":71,"doi":72,"paper":73,"created_at":92},4822,"Hybrid transformer and CNN for hyperspectral image classification with spectral band selection","https:\u002F\u002Fdoi.org\u002F10.1038\u002Fs41598-026-72102-z","Abstract Hyperspectral images (HSI) play a vital role in remote sensing applications such as surveillance and precision agriculture, yet they face challenges including limited labeled samples and class imbalance. This paper presents a new method for HSI classification by integrating spectral band selection with advanced deep learning techniques. First, we propose an optimized band selection method alongside data augmentation using crop-and-resize transformations to enhance feature diversity. Second, we introduce a hybrid convolution-transformer framework that combines Vision Transformer (ViT) and 2D-CNN architectures, supplemented with a sequence aggregation layer to mitigate overfitting on small datasets. This hybrid model effectively captures both spatial and spectral features while preserving critical spectral details. Finally, we employ a support vector machine for robust classification. Extensive experiments on three widely used HSI datasets demonstrate that our proposed ViT-CNN-BS framework achieves superior classification accuracy compared to state-of-the-art few-shot learning methods and traditional approaches. Beyond achieving higher accuracy, our approach also reduces computational complexity, making it suitable for resource-constrained applications. The findings highlight the potential of hybrid deep learning frameworks for advancing remote sensing analytics, with possible extensions to environmental monitoring, disaster management, and precision agriculture.","高光谱图像（HSI）在监视和精准农业等遥感应用中发挥着重要作用，但仍面临标注样本有限和类别不平衡等挑战。本文提出了一种将光谱波段选择与先进深度学习技术相结合的高光谱图像分类新方法。首先，我们提出了一种优化的波段选择方法，并利用裁剪与缩放变换进行数据增强，以提升特征多样性。其次，我们引入了一种混合卷积-Transformer框架，将视觉Transformer（ViT）与二维卷积神经网络（2D-CNN）架构相结合，并辅以序列聚合层以缓解小数据集上的过拟合问题。该混合模型在保留关键光谱细节的同时，有效捕捉了空间和光谱特征。最后，我们采用支持向量机进行稳健分类。在三个广泛使用的高光谱图像数据集上进行的大量实验表明，我们提出的ViT-CNN-BS框架相比最先进的少样本学习方法和传统方法，实现了更优的分类精度。除了达到更高的精度外，我们的方法还降低了计算复杂度，使其适用于资源受限的应用场景。研究结果凸显了混合深度学习框架在推动遥感分析方面的潜力，并可扩展应用于环境监测、灾害管理和精准农业等领域。",null,"Scientific Reports","2026-10-12T00:00:00Z","论文",10,false,73,{"impact":17,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":23},12,21,17,14,9,1,"高光谱遥感分类新方法，方法新颖且面向精准农业等应用，但属实验室数据集验证，产业落地影响有限。",[25],{"name":10,"url":6},[27,28,29,30,31],"农业人工智能","精准农业","遥感","作物分类","高光谱影像",{"code":33,"slug":34,"name":35,"path":36,"parent_code":37},"I3","remote-sensing","遥感与农情监测","\u002Ftopic\u002Fdigital-agriculture\u002Fremote-sensing","I",[],{"status":40,"locked":14,"version":41,"method":42,"evidence":43},"classified","2026-10-22.2","rules",{"candidates":44,"agricultural_evidence":47},[45,53,60],{"code":33,"score":46,"supported":47,"title":48,"abstract":50,"tags":52},19,true,[49],"hyperspectral",[29,51,49],"remote sensing",[30,29],{"code":54,"score":55,"supported":47,"title":56,"abstract":57,"tags":59},"I1",7,[],[28,58],"precision agriculture",[28],{"code":61,"score":55,"supported":47,"title":62,"abstract":63,"tags":66},"I2",[],[64,65],"深度学习","deep learning",[27],[68,69],"高光谱影像 分类 ViT-CNN","光谱波段选择 精准农业","高光谱影像分类ViT-CNN-4822",2,"10.1038\u002Fs41598-026-72102-z",{"doi":72,"openalex_id":74,"authors":75,"venue":10,"cited_by_count":84,"oa_url":6,"card":85,"direction":89,"ingested_from":91},"W7221518213",[76,78,81],{"name":77,"orcid":9},"Eman N. Abdelhafez",{"name":79,"orcid":80},"Ahmed Hagag","https:\u002F\u002Forcid.org\u002F0000-0003-2631-1846",{"name":82,"orcid":83},"Tamer Ahmed Abassy","https:\u002F\u002Forcid.org\u002F0000-0002-1960-7748",0,{"tldr":86,"method":87,"finding":88,"direction":89,"opportunity":90},"提出结合波段选择与ViT-CNN混合框架的高光谱图像分类方法，提升小样本分类精度。","优化波段选择、裁剪缩放数据增强、ViT与2D-CNN混合模型及SVM分类。","在三个高光谱数据集上分类精度优于少样本学习与传统方法，且计算复杂度更低。","农业遥感与作物表型","可探索轻量化混合模型在资源受限的田间实时作物分类与病害监测中的部署。","openalex","2026-10-13T23:30:23.758699Z",{"total":94,"page":22,"page_size":94,"items":95},6,[96,183,245,346,411,499],{"id":97,"title":98,"url":99,"summary":100,"summary_zh":101,"content":9,"source_name":102,"source_url":99,"published_at":103,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":104,"score_detail":105,"sources":109,"tags":111,"primary_topic":114,"secondary_topics":119,"classification":121,"search_phrases":148,"slug":151,"view_count":152,"doi":153,"paper":154,"created_at":182},4356,"Dual‐Branch Network With Enhanced Feature Fusion for Improved Rice Classification Accuracy","https:\u002F\u002Fdoi.org\u002F10.1002\u002Fsae2.70218","ABSTRACT Accurate mapping of rice, one of the world's most important food crops, is critical for monitoring its growth, estimating yields, and supporting global food security. Most deep learning methods for crop classification rely on multi‐temporal optical remote sensing data. However, it is difficult to obtain sufficient ground‐truth samples for time‐series deep learning models through conducting large‐scale field surveys, and persistent cloud and rainfall often limit optical observations during the rice‐growing seasons in subtropical regions. Rice classification based solely on time‐series optical images may therefore be less accurate and reliable. Here, we developed two dual‐branch deep learning networks for effective rice identification: one combining a long short‐term memory network running in parallel with a multi‐layer perceptron (LSTM + MLP), and the other replacing the LSTM with a one‐dimensional convolutional neural network (Conv1d + MLP). These models process temporal sequences in parallel with auxiliary data through the MLP, allowing them to automatically extract phenology information while incorporating auxiliary variables such as texture features, digital elevation model (DEM) data, and the standard deviation of the vegetation index time series. We evaluated these models in a representative county in the subtropical region of China, and compared the accuracies with standalone LSTM, Conv1d, MLP, and random forest (RF) models. The results revealed that LSTM + MLP had the highest accuracy, with an F1‐score of 95.5%, which can automatically extract rice growth patterns and successfully identify the key phenological information of flood signals during the rice transplanting period. By incorporating auxiliary variables, the model achieved an average accuracy improvement of 8.1% relative to the models using only temporal features. The proposed method provides a useful approach for rice mapping in subtropical regions.","准确绘制水稻这一全球最重要的粮食作物之一，对于监测其生长、估算产量以及保障全球粮食安全至关重要。大多数作物分类的深度学习方法依赖于多时相光学遥感数据。然而，通过开展大规模实地调查来获取足够的地面真值样本以用于时间序列深度学习模型十分困难，而且亚热带地区水稻生长季期间持续的云层和降雨往往限制了光学观测。因此，仅基于时间序列光学图像的水稻分类可能精度较低且可靠性不足。本研究开发了两种双分支深度学习网络以实现有效的水稻识别：一种将长短期记忆网络与多层感知机并行运行（LSTM + MLP），另一种则用一维卷积神经网络替代LSTM（Conv1d + MLP）。这些模型通过MLP将时间序列与辅助数据并行处理，使其能够自动提取物候信息，同时纳入纹理特征、数字高程模型（DEM）数据以及植被指数时间序列标准差等辅助变量。我们在中国亚热带地区一个代表性县对这些模型进行了评估，并将精度与单独的LSTM、Conv1d、MLP和随机森林（RF）模型进行了比较。结果表明，LSTM + MLP精度最高，F1分数达到95.5%，能够自动提取水稻生长模式，并成功识别水稻移栽期洪水信号的关键物候信息。通过纳入辅助变量，该模型相对于仅使用时间特征的模型平均精度提高了8.1%。所提出的方法为亚热带地区水稻制图提供了一种有效途径。","Journal of Sustainable Agriculture and Environment","2026-10-06T00:00:00Z",75,{"impact":106,"substance":18,"depth":19,"authority":107,"freshness":21,"relevant":22,"comment":108},15,13,"提出双分支深度学习网络融合时序与辅助数据，水稻识别F1达95.5%，方法新颖且结论可靠，对亚热带多云雨区作物遥感制图有实用价值。",[110],{"name":102,"url":99},[112,27,113,29,30],"智慧农业","水稻",{"code":115,"slug":116,"name":113,"path":117,"parent_code":118},"A1","rice","\u002Ftopic\u002Ffield-crops\u002Frice","A",[120],{"code":33,"slug":34,"name":35,"path":36,"parent_code":37},{"status":40,"locked":14,"version":41,"method":42,"evidence":122},{"candidates":123,"agricultural_evidence":47},[124,128,133,137,144],{"code":115,"score":106,"supported":47,"title":125,"abstract":126,"tags":127},[116],[113,116],[113],{"code":33,"score":129,"supported":47,"title":130,"abstract":131,"tags":132},8,[],[29,51],[30,29],{"code":61,"score":55,"supported":47,"title":134,"abstract":135,"tags":136},[],[64,65],[27],{"code":138,"score":94,"supported":47,"title":139,"abstract":140,"tags":143},"A8",[],[141,142],"粮食安全","food security",[],{"code":54,"score":22,"supported":14,"title":145,"abstract":146,"tags":147},[],[],[112],[149,150],"水稻 遥感 分类 深度学习","亚热带 水稻 制图","水稻遥感分类深度学习-4356",4,"10.1002\u002Fsae2.70218",{"doi":153,"openalex_id":155,"authors":156,"venue":102,"cited_by_count":84,"oa_url":176,"card":177,"direction":89,"ingested_from":91},"W7220691059",[157,159,162,165,167,170,173],{"name":158,"orcid":9},"Rongfei Ma",{"name":160,"orcid":161},"Wenjiao Shi","https:\u002F\u002Forcid.org\u002F0000-0001-6199-9733",{"name":163,"orcid":164},"Minglei Wang","https:\u002F\u002Forcid.org\u002F0000-0002-4174-9224",{"name":166,"orcid":9},"Xiwei Chen",{"name":168,"orcid":169},"Chenchen Zhang","https:\u002F\u002Forcid.org\u002F0000-0003-0458-6232",{"name":171,"orcid":172},"Baihong Pan","https:\u002F\u002Forcid.org\u002F0009-0004-1100-6798",{"name":174,"orcid":175},"Xiangming Xiao","https:\u002F\u002Forcid.org\u002F0000-0003-0956-7428","https:\u002F\u002Fonlinelibrary.wiley.com\u002Fdoi\u002Fpdfdirect\u002F10.1002\u002Fsae2.70218",{"tldr":178,"method":179,"finding":180,"direction":89,"opportunity":181},"提出双分支深度学习网络融合时序与辅助数据，提升亚热带水稻分类精度。","LSTM\u002FConv1d与MLP并行双分支网络，融合时序植被指数、纹理、DEM等辅","LSTM+MLP模型F1达95.5%，辅助变量使精度平均提升8.1%，能识别插秧期淹水信号。","可探索多源遥感与气象数据融合，解决多云雨地区时序缺失下的作物精细制图问题。","2026-10-07T23:30:25.945309Z",{"id":184,"title":185,"url":186,"summary":187,"summary_zh":188,"content":9,"source_name":189,"source_url":186,"published_at":190,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":191,"score_detail":192,"sources":194,"tags":196,"primary_topic":198,"secondary_topics":203,"classification":205,"search_phrases":222,"slug":225,"view_count":226,"doi":227,"paper":228,"created_at":244},3179,"Multi-features and Fuzzy with Residual Lenet for Cotton Classification Using Remote Sensing Image","https:\u002F\u002Fdoi.org\u002F10.1007\u002Fs12524-026-02541-8","Multi-features and Fuzzy with Residual Lenet for Cotton Classification Using Remote Sensing Image。Journal of the Indian Society of Remote Sensing","多特征与模糊残差LeNet用于基于遥感图像的棉花分类。《印度遥感学会杂志》","Journal of the Indian Society of Remote Sensing","2026-09-21T00:00:00Z",58,{"impact":129,"substance":20,"depth":106,"authority":17,"freshness":21,"relevant":22,"comment":193},"方法类论文，提出多特征与模糊结合残差LeNet用于遥感棉花分类，专业价值尚可但影响面窄，适合作为技术参考而非每日精选头条。",[195],{"name":189,"url":186},[27,197,29,30],"棉花",{"code":199,"slug":200,"name":201,"path":202,"parent_code":118},"A6","cash-crops","棉麻糖与特色经作","\u002Ftopic\u002Ffield-crops\u002Fcash-crops",[204],{"code":33,"slug":34,"name":35,"path":36,"parent_code":37},{"status":40,"locked":14,"version":41,"method":42,"evidence":206},{"candidates":207,"agricultural_evidence":47},[208,213,218],{"code":33,"score":209,"supported":47,"title":210,"abstract":211,"tags":212},16,[51],[29,51],[30,29],{"code":199,"score":106,"supported":47,"title":214,"abstract":216,"tags":217},[215],"cotton",[197,215],[197],{"code":61,"score":22,"supported":14,"title":219,"abstract":220,"tags":221},[],[],[27],[223,224],"Residual LeNet 棉花 遥感分类","农业人工智能 作物分类 棉花 遥感","ResidualLeNet棉花遥感分类-3179",5,"10.1007\u002Fs12524-026-02541-8",{"doi":227,"openalex_id":229,"authors":230,"venue":189,"cited_by_count":84,"oa_url":9,"card":239,"direction":89,"ingested_from":91},"W7213944040",[231,233,235,237],{"name":232,"orcid":9},"G. Suresh",{"name":234,"orcid":9},"G. Bhuvaneswari",{"name":236,"orcid":9},"G. Manikandan",{"name":238,"orcid":9},"M. Robinson Joel",{"tldr":240,"method":241,"finding":242,"direction":89,"opportunity":243},"提出多特征与模糊结合残差LeNet的遥感影像棉花分类方法。","多特征提取、模糊逻辑与残差LeNet网络，用于遥感图像分类。","该方法能有效提升棉花分类精度，验证了深度残差网络结合模糊特征的可行性。","可探索多特征与模糊深度学习在其他作物遥感分类中的泛化能力及轻量化部署。","2026-09-22T23:30:23.811658Z",{"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":257,"tags":259,"primary_topic":261,"secondary_topics":267,"classification":273,"search_phrases":309,"slug":312,"view_count":94,"doi":313,"paper":314,"created_at":345},2947,"AI-enabled UAV-based Soil Organic Carbon Mapping in Arid Environments: A Pilot Study Protocol","https:\u002F\u002Fdoi.org\u002F10.2174\u002F0118743315495282260915110324","Introduction Soil organic carbon (SOC) is an important indicator of soil health, agricultural productivity, and carbon sequestration potential. However, accurate and scalable SOC mapping in arid environments is constrained by high spatial heterogeneity and the limitations of conventional soil sampling. This study aims to develop a standardized UAV-enabled framework for high-resolution SOC mapping in arid agricultural environments. Methods A pilot-study protocol integrating UAV-based hyperspectral remote sensing with artificial intelligence and machine learning was developed. The workflow encompasses study-site selection, ground-reference sampling, UAV hyperspectral data acquisition, radiometric and geometric preprocessing, spectral feature extraction and selection, machine-learning model development, validation, uncertainty assessment, and performance evaluation using R 2 , RMSE, and MAE. The protocol also incorporates assessment of environmental confounders, including soil moisture, surface roughness, and crop residues. Results The resulting framework provides a systematic and reproducible workflow for UAV-based SOC estimation, integrating field observations, hyperspectral features, predictive modelling, and uncertainty assessment. It establishes defined procedures for evaluating model robustness and transferability across varying field conditions. Discussion The framework addresses an important methodological gap in UAV-enabled SOC mapping by integrating remote sensing and AI within a standardized pilot-study design. Its emphasis on environmental confounders and uncertainty assessment can improve the reliability and comparability of SOC mapping studies. However, field validation across diverse arid environments remains necessary. Conclusion The proposed protocol provides a practical foundation for reproducible SOC mapping and subsequent field validation, supporting precision agriculture, sustainable soil management, and carbon monitoring, reporting, and verification (MRV) in arid regions.","引言 土壤有机碳（SOC）是衡量土壤健康、农业生产力及碳固存潜力的重要指标。然而，干旱环境中高空间异质性和传统土壤采样的局限性制约了准确且可扩展的SOC制图。本研究旨在开发一个标准化的无人机（UAV）框架，用于干旱农业环境中的高分辨率SOC制图。方法 开发了一套整合无人机高光谱遥感与人工智能及机器学习的试点研究方案。该工作流程涵盖研究地点选择、地面参考采样、无人机高光谱数据采集、辐射与几何预处理、光谱特征提取与选择、机器学习模型开发、验证、不确定性评估，以及使用R²、RMSE和MAE进行的性能评价。该方案还包括对环境混杂因素的评估，包括土壤水分、地表粗糙度和作物残茬。结果 所构建的框架为基于无人机的SOC估算提供了系统且可重复的工作流程，整合了野外观测、高光谱特征、预测建模和不确定性评估。它建立了明确的程序，用于评估模型在不同田间条件下的稳健性和可迁移性。讨论 该框架通过将遥感与人工智能整合于标准化的试点研究设计中，填补了无人机SOC制图领域的重要方法学空白。其对环境混杂因素和不确定性评估的重视，可提高SOC制图研究的可靠性和可比性。然而，仍需在不同干旱环境中进行田间验证。结论 所提出的方案为可重复的SOC制图及后续田间验证提供了实用基础，支持干旱地区的精准农业、可持续土壤管理以及碳监测、报告与核查（MRV）。","The Open Agriculture Journal","2026-09-18T00:00:00Z",67,{"impact":17,"substance":255,"depth":209,"authority":107,"freshness":129,"relevant":22,"comment":256},18,"提出无人机高光谱结合AI的干旱区土壤有机碳制图标准化方案，方法框架清晰但尚属试点协议、缺乏实地验证，具备一定参考价值。",[258],{"name":251,"url":248},[112,27,28,29,260],"土壤碳汇",{"code":262,"slug":263,"name":264,"path":265,"parent_code":266},"H2","soil","土壤健康与改良","\u002Ftopic\u002Fenvironment\u002Fsoil","H",[268,272],{"code":61,"slug":269,"name":270,"path":271,"parent_code":37},"agricultural-ai","农业AI与大模型","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-ai",{"code":33,"slug":34,"name":35,"path":36,"parent_code":37},{"status":40,"locked":14,"version":41,"method":42,"evidence":274},{"candidates":275,"agricultural_evidence":47},[276,281,288,292,296,302],{"code":262,"score":20,"supported":47,"title":277,"abstract":278,"tags":280},[263],[279,263],"土壤",[],{"code":61,"score":13,"supported":47,"title":282,"abstract":283,"tags":287},[],[284,285,286],"机器学习","machine learning","artificial intelligence",[27],{"code":33,"score":13,"supported":47,"title":289,"abstract":290,"tags":291},[],[29,51,49],[29],{"code":54,"score":129,"supported":47,"title":293,"abstract":294,"tags":295},[],[28,58],[112,28],{"code":297,"score":152,"supported":47,"title":298,"abstract":299,"tags":301},"H4",[],[300],"carbon sequestration",[260],{"code":303,"score":304,"supported":47,"title":305,"abstract":306,"tags":308},"H7",3,[],[307],"干旱",[],[310,311],"无人机 土壤有机碳 制图","AI 高光谱 干旱农业","无人机土壤有机碳制图-2947","10.2174\u002F0118743315495282260915110324",{"doi":313,"openalex_id":315,"authors":316,"venue":251,"cited_by_count":84,"oa_url":248,"card":340,"direction":89,"ingested_from":91},"W7213561504",[317,320,323,326,329,332,334,336,338],{"name":318,"orcid":319},"Moath Awawdeh","https:\u002F\u002Forcid.org\u002F0000-0003-1404-6782",{"name":321,"orcid":322},"Irfan Ahmed","https:\u002F\u002Forcid.org\u002F0000-0002-2172-4177",{"name":324,"orcid":325},"Anees Bashir","https:\u002F\u002Forcid.org\u002F0000-0002-4668-6592",{"name":327,"orcid":328},"Tarig Faisal","https:\u002F\u002Forcid.org\u002F0000-0001-6451-7576",{"name":330,"orcid":331},"Nicky Rahmana Putra","https:\u002F\u002Forcid.org\u002F0000-0003-4886-496X",{"name":333,"orcid":9},"Almaha Jamal",{"name":335,"orcid":9},"Afra Rashed",{"name":337,"orcid":9},"Hamda Yousif",{"name":339,"orcid":9},"Sarah Sadeq",{"tldr":341,"method":342,"finding":343,"direction":89,"opportunity":344},"提出一套无人机高光谱结合AI的干旱区土壤有机碳制图标准化试点方案。","无人机高光谱遥感、地面采样、光谱特征选择与机器学习建模，用R²、RMSE、MAE","构建了可复现的SOC估算流程，并纳入环境混杂因素与不确定性评估。","可在多干旱区开展跨区域验证，探索模型迁移性与不确定性量化方法。","2026-09-19T23:30:33.273156Z",{"id":347,"title":348,"url":349,"summary":350,"summary_zh":351,"content":9,"source_name":352,"source_url":349,"published_at":252,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":353,"score_detail":354,"sources":357,"tags":359,"primary_topic":361,"secondary_topics":362,"classification":364,"search_phrases":385,"slug":388,"view_count":22,"doi":389,"paper":390,"created_at":410},2942,"Deep learning-driven multisource remote sensing image fusion: Advances, challenges, and future directions","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.engappai.2026.116319","Multisource remote sensing image fusion has become an important solution to a long-standing limitation in Earth observation: individual sensors rarely provide high spatial detail, rich spectral information, reliable structural sensitivity, and frequent temporal coverage at the same time. This review examines how deep learning and artificial intelligence are being used to integrate multispectral, hyperspectral, panchromatic, optical, and synthetic aperture radar imagery for more reliable interpretation of complex ground scenes. It provides a technical synthesis of convolutional neural networks, autoencoders, generative adversarial networks, transformer architectures, diffusion models, and hybrid model driven approaches, with attention to their fusion mechanisms, reconstruction behavior, computational demand, and suitability for operational use. Applications include land cover mapping, precision agriculture, environmental monitoring, urban analysis, disaster assessment, and defense related interpretation. Rather than treating each fusion task separately, this review connects sensor heterogeneity, spatial and spectral resolution trade offs, radiometric correction, geometric correction, registration, noise reduction, and fusion level design within a single framework. The analysis indicates that convolutional models remain effective for stable local detail recovery, adversarial learning can improve visual sharpness but may introduce spectral distortion, transformer models better capture long range spatial and spectral relationships, and diffusion models offer refined reconstruction at greater computational cost. The review further identifies open challenges involving misregistration, spectral bias, limited labeled data, weak generalization across sensors, high memory requirements, and limited interpretability. Future progress should prioritize sensor aware learning, self supervised training, uncertainty aware evaluation, lightweight deployment, and application oriented benchmarks to improve reliability in operational Earth observation.","多源遥感图像融合已成为解决地球观测领域一个长期局限的重要方案：单一传感器很少能够同时提供高空间细节、丰富光谱信息、可靠的结构敏感性以及频繁的时间覆盖。本文综述了如何利用深度学习和人工智能整合多光谱、高光谱、全色、光学和合成孔径雷达（synthetic aperture radar, SAR）影像，以更可靠地解译复杂地表场景。文章对卷积神经网络、自编码器、生成对抗网络、Transformer架构、扩散模型以及混合模型驱动方法进行了技术综合，重点关注其融合机制、重建行为、计算需求以及业务化适用性。应用领域包括土地覆盖制图、精准农业、环境监测、城市分析、灾害评估和国防相关解译。本文并非将每种融合任务分开处理，而是在一个统一框架内将传感器异质性、空间与光谱分辨率权衡、辐射校正、几何校正、配准、降噪和融合层级设计联系起来。分析表明，卷积模型在稳定的局部细节恢复方面仍然有效，对抗学习可以提升视觉锐度但可能引入光谱失真，Transformer模型能更好地捕捉长程空间与光谱关系，而扩散模型以更高的计算成本提供精细重建。本文进一步指出了涉及配准误差、光谱偏差、标注数据有限、跨传感器泛化能力弱、高内存需求以及可解释性有限等开放挑战。未来的进展应优先关注传感器感知学习、自监督训练、不确定性感知评估、轻量化部署以及面向应用的基准测试，以提高业务化地球观测的可靠性。","Engineering Applications of Artificial Intelligence",82,{"impact":255,"substance":355,"depth":46,"authority":20,"freshness":21,"relevant":22,"comment":356},22,"发表于核心期刊的综述，系统梳理深度学习多源遥感融合的方法、应用与挑战，对农业遥感与精准农业有直接参考价值，时效性强，值得进入每日精选。",[358],{"name":352,"url":349},[27,64,28,29,360],"多源数据融合",{"code":61,"slug":269,"name":270,"path":271,"parent_code":37},[363],{"code":33,"slug":34,"name":35,"path":36,"parent_code":37},{"status":40,"locked":14,"version":41,"method":42,"evidence":365},{"candidates":366,"agricultural_evidence":47},[367,371,376,380],{"code":61,"score":46,"supported":47,"title":368,"abstract":369,"tags":370},[65],[64,65,286],[27,64],{"code":33,"score":255,"supported":47,"title":372,"abstract":373,"tags":375},[51],[29,51,374,49],"multispectral",[29],{"code":54,"score":55,"supported":47,"title":377,"abstract":378,"tags":379},[],[28,58],[28],{"code":381,"score":22,"supported":14,"title":382,"abstract":383,"tags":384},"I5",[],[],[360],[386,387],"多源遥感 图像融合 深度学习","农业人工智能 多源数据融合 深度学习 精准农业","多源遥感图像融合深度学习-2942","10.1016\u002Fj.engappai.2026.116319",{"doi":389,"openalex_id":391,"authors":392,"venue":352,"cited_by_count":84,"oa_url":349,"card":405,"direction":89,"ingested_from":91},"W7213544281",[393,396,399,402],{"name":394,"orcid":395},"Shahid Karim","https:\u002F\u002Forcid.org\u002F0000-0001-9986-5052",{"name":397,"orcid":398},"Akeel Qadir","https:\u002F\u002Forcid.org\u002F0000-0003-0358-6505",{"name":400,"orcid":401},"Asif Ali Laghari","https:\u002F\u002Forcid.org\u002F0000-0001-5831-5943",{"name":403,"orcid":404},"Irfana Bibi","https:\u002F\u002Forcid.org\u002F0000-0003-2794-504X",{"tldr":406,"method":407,"finding":408,"direction":89,"opportunity":409},"综述深度学习多源遥感图像融合方法、挑战与未来方向。","综述CNN、GAN、Transformer、扩散模型等融合机制与重建行为。","CNN擅局部细节，GAN易谱失真，Transformer长程关系强，扩散模型精度高但算力大。","面向农业的轻量、自监督、不确定性感知融合与基准数据集构建。","2026-09-19T23:30:32.767384Z",{"id":412,"title":413,"url":414,"summary":415,"summary_zh":416,"content":9,"source_name":417,"source_url":414,"published_at":418,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":353,"score_detail":419,"sources":421,"tags":423,"primary_topic":426,"secondary_topics":427,"classification":428,"search_phrases":455,"slug":458,"view_count":22,"doi":459,"paper":460,"created_at":498},2786,"SAF-CropNet: Spatially adaptive fusion of SAR and optical imagery for semantic segmentation in operational cropland mapping across regions","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.isprsjprs.2026.08.029","Accurate cropland mapping from remote sensing imagery is essential for agricultural monitoring, food security assessment, and land resource management. Although optical and synthetic aperture radar (SAR) observations provide complementary information for cropland extraction, their local contributions vary across imaging conditions, land-cover backgrounds, parcel morphologies, and regional agricultural systems. This variability limits direct concatenation and static fusion, as modality-specific errors may propagate into segmentation outputs in the absence of spatially adaptive weighting. This study proposes SAF-CropNet, a spatially adaptive SAR–optical fusion framework for binary cropland semantic segmentation. Specifically, SAF-CropNet first extracts modality-specific SAR structural features and optical spectral-textural features through separate encoder branches. The Paired Reciprocal Inter-modal Selective Modulator enhances complementary SAR–optical interactions, while the Contextual State-space Modeling module captures long-range parcel organization. A spatially adaptive fusion head then estimates per-pixel contribution weights for SAR, optical, and contextual features before decoding. Evaluation was conducted on a seven-area SAR–optical benchmark spanning China, Germany, and France. The benchmark integrates GF1\u002FGF3 and Sentinel-1\u002F2 imagery with reference labels derived from field surveys and RapidCrops products and covers fragmented smallholder systems, water-rich and peri-urban mosaics, and large mechanized cropland. Within-region five-fold validation shows that SAF-CropNet achieves an average F1 score of 0.8550 and an mIoU of 0.8068, outperforming the evaluated segmentation baselines while remaining lightweight, with 3.38 M parameters and 6.21 GFLOPs for 256 × 256 SAR–optical inputs. In leave-one-area-out transfer experiments, stratified few-shot adaptation increased the average F1 score from 0.4986 under zero-shot transfer to 0.7698. Grouped cross-continental and cross-sensor experiments further showed that direct transfer remains constrained under compound shifts in sensor characteristics, spatial resolution, landscape structure, and reference-label conventions. These findings indicate that SAF-CropNet combines strong within-region segmentation accuracy and computational efficiency with substantial gains from limited target-domain adaptation, while direct zero-shot generalization under severe compound domain shifts remains limited.","从遥感影像中准确提取耕地信息对农业监测、粮食安全评估和土地资源管理至关重要。尽管光学与合成孔径雷达（SAR）观测为耕地提取提供了互补信息，但其局部贡献会随成像条件、地表覆盖背景、地块形态和区域农业系统的不同而变化。这种变异性限制了直接拼接和静态融合的效果，因为在缺乏空间自适应加权的情况下，模态特有的误差可能传播至分割输出中。本研究提出SAF-CropNet，一种用于二分类耕地语义分割的空间自适应SAR–光学融合框架。具体而言，SAF-CropNet首先通过独立的编码器分支提取模态特有的SAR结构特征和光学光谱-纹理特征。成对互反模态间选择性调制器增强SAR与光学之间的互补交互，而上下文状态空间建模模块则捕获长距离地块组织信息。随后，空间自适应融合头在解码前估计SAR、光学和上下文特征的逐像素贡献权重。评估在一个涵盖中国、德国和法国的七区域SAR–光学基准上进行。该基准整合了GF1\u002FGF3和Sentinel-1\u002F2影像，参考标签来源于实地调查和RapidCrops产品，覆盖破碎化小农系统、富水与城郊镶嵌景观以及大规模机械化耕地。区域内五折验证表明，SAF-CropNet平均F1分数达到0.8550，mIoU为0.8068，优于所评估的分割基线，同时保持轻量级，对于256 × 256的SAR–光学输入仅需3.38 M参数和6.21 GFLOPs。在留一区域迁移实验中，分层少样本自适应将平均F1分数从零样本迁移下的0.4986提升至0.7698。分组跨大陆和跨传感器实验进一步表明，在传感器特性、空间分辨率、景观结构和参考标签惯例的复合变化下，直接迁移仍然受到限制。这些发现表明，SAF-CropNet兼具较强的区域内分割精度和计算效率，并能从有限的目标域自适应中获得显著增益，而在严重复合域偏移下的直接零样本泛化能力仍然有限。","ISPRS Journal of Photogrammetry and Remote Sensing","2026-09-17T00:00:00Z",{"impact":255,"substance":355,"depth":255,"authority":20,"freshness":13,"relevant":22,"comment":420},"提出空间自适应SAR-光学融合分割框架，跨七区域基准验证，兼顾精度与轻量化，对耕地遥感制图有实质方法贡献。",[422],{"name":417,"url":414},[27,29,424,30,425],"耕地监测","SAR",{"code":33,"slug":34,"name":35,"path":36,"parent_code":37},[],{"status":40,"locked":14,"version":41,"method":42,"evidence":429},{"candidates":430,"agricultural_evidence":47},[431,435,439,445,451],{"code":33,"score":129,"supported":47,"title":432,"abstract":433,"tags":434},[],[29,51],[30,29],{"code":138,"score":94,"supported":47,"title":436,"abstract":437,"tags":438},[],[141,142],[],{"code":440,"score":304,"supported":47,"title":441,"abstract":442,"tags":444},"F5",[],[443],"机械化",[],{"code":446,"score":304,"supported":47,"title":447,"abstract":448,"tags":450},"G2",[],[449],"smallholder",[],{"code":61,"score":22,"supported":14,"title":452,"abstract":453,"tags":454},[],[],[27],[456,457],"农业人工智能 作物分类 耕地监测 遥感","农业人工智能 作物分类","农业人工智能作物分类耕地监测遥感-2786","10.1016\u002Fj.isprsjprs.2026.08.029",{"doi":459,"openalex_id":461,"authors":462,"venue":417,"cited_by_count":84,"oa_url":414,"card":493,"direction":89,"ingested_from":91},"W7213463520",[463,466,468,471,474,477,480,482,484,486,488,490],{"name":464,"orcid":465},"minghui chang","https:\u002F\u002Forcid.org\u002F0009-0008-7954-5308",{"name":467,"orcid":9},"Shuaifeng Peng",{"name":469,"orcid":470},"Tao Xu","https:\u002F\u002Forcid.org\u002F0000-0002-7855-4199",{"name":472,"orcid":473},"Yi Yuan","https:\u002F\u002Forcid.org\u002F0009-0006-9481-4938",{"name":475,"orcid":476},"洋一 馬目","https:\u002F\u002Forcid.org\u002F0000-0003-0226-5558",{"name":478,"orcid":479},"Jie Bai","https:\u002F\u002Forcid.org\u002F0000-0003-2426-2358",{"name":481,"orcid":9},"Fugui Luo",{"name":483,"orcid":9},"Jingyu Zhang",{"name":485,"orcid":9},"Xiaoyu Xiao",{"name":487,"orcid":9},"Yu Mu",{"name":489,"orcid":9},"Yong Wang",{"name":491,"orcid":492},"Shihua Li","https:\u002F\u002Forcid.org\u002F0000-0003-4807-5012",{"tldr":494,"method":495,"finding":496,"direction":89,"opportunity":497},"提出SAF-CropNet，用空间自适应融合SAR与光学影像做跨区域耕地语义分割。","双分支编码器提取SAR结构\u002F光学纹理特征，结合互模态调制与状态空间建模，在七区域","区域内F1达0.8550且轻量；零样本迁移F1仅0.4986，少样本适配可升至0.7698。","复合域偏移下零样本泛化差，可研究跨传感器、跨区域的无监督域自适应融合分割方法。","2026-09-17T23:30:34.424579Z",{"id":500,"title":501,"url":502,"summary":503,"summary_zh":504,"content":9,"source_name":352,"source_url":502,"published_at":505,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":506,"score_detail":507,"sources":509,"tags":511,"primary_topic":513,"secondary_topics":514,"classification":515,"search_phrases":530,"slug":533,"view_count":71,"doi":534,"paper":535,"created_at":562},2165,"Multi-grained image-text retrieval in farmland remote sensing via multimodal large language models","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.engappai.2026.116113","Low-altitude farmland remote sensing (FRS) has become an important tool for precision agricultural monitoring. However, existing FRS methods are mainly designed for segmentation with predefined categories, making them insufficient for semantic understanding and unable to support flexible natural-language queries. To address this limitation, we introduce an artificial intelligence-based task, FRS cross-modal retrieval, which aims to retrieve target farmland regions from large-scale remote sensing images using textual descriptions. To support this task, we construct the first Farmland Remote Sensing Cross-Modal Retrieval (FRS-CMR) dataset, containing 67,806 image-text pairs generated through multimodal large language models and expert refinement. Based on FRS-CMR, we propose a Farmland-oriented Multi-Grained Retrieval (FarmMGR) framework for image-text alignment. FarmMGR jointly exploits global scene semantics and local region-level cues to capture both large-scale landscape patterns and fine-grained farmland characteristics. To further improve semantic alignment, we design Prototype-Based Semantic Alignment (PBSA) to alleviate the soft-positive sample problem, Global Semantics Hierarchical Feature Integration (GSHFI) to adaptively fuse global and local representations, and Instance-Level Semantic Relation Modeling (ISRM) to capture relational dependencies among farmland instances. In addition, Memory-Enhanced Cross-Modal Contrastive Learning (MECCL) enlarges the negative sample pool and improves representation discrimination. Extensive experiments on FRS-CMR show that FarmMGR achieves a mean recall of 28.57%, outperforming existing image-text retrieval methods and demonstrating its effectiveness for language-driven farmland remote sensing retrieval.","低空农田遥感(FRS)已成为精准农业监测的重要工具。然而，现有农田遥感方法主要针对预定义类别的分割任务设计，难以进行语义理解，也无法支持灵活的自然语言查询。为解决这一局限，我们引入了一项基于人工智能的任务——农田遥感跨模态检索，旨在利用文本描述从大规模遥感图像中检索目标农田区域。为支撑该任务，我们构建了首个农田遥感跨模态检索(FRS-CMR)数据集，包含67,806个通过多模态大语言模型生成并经专家精炼的图像-文本对。基于FRS-CMR，我们提出了面向农田的多粒度检索(FarmMGR)框架以实现图像-文本对齐。FarmMGR联合利用全局场景语义和局部区域级线索，以同时捕获大尺度景观格局和细粒度农田特征。为进一步提升语义对齐效果，我们设计了基于原型的语义对齐(PBSA)以缓解软正样本问题，全局语义层次特征融合(GSHFI)以自适应融合全局与局部表示，以及实例级语义关系建模(ISRM)以捕获农田实例间的关系依赖。此外，记忆增强跨模态对比学习(MECCL)扩大了负样本池并提升了表示判别能力。在FRS-CMR上的大量实验表明，FarmMGR取得了28.57%的平均召回率，优于现有图像-文本检索方法，证明了其在语言驱动农田遥感检索中的有效性。","2026-09-10T00:00:00Z",81,{"impact":255,"substance":355,"depth":255,"authority":20,"freshness":21,"relevant":22,"comment":508},"构建首个农田遥感跨模态检索数据集并提出多粒度对齐框架，方法新颖、数据规模可观，对语言驱动的精准农业监测有实质推动。",[510],{"name":352,"url":502},[112,27,28,29,512],"多模态大模型",{"code":33,"slug":34,"name":35,"path":36,"parent_code":37},[],{"status":40,"locked":14,"version":41,"method":42,"evidence":516},{"candidates":517,"agricultural_evidence":47},[518,522,526],{"code":33,"score":106,"supported":47,"title":519,"abstract":520,"tags":521},[51],[29,51],[29],{"code":54,"score":226,"supported":47,"title":523,"abstract":524,"tags":525},[],[28],[112,28],{"code":61,"score":226,"supported":47,"title":527,"abstract":528,"tags":529},[],[286],[27,512],[531,532],"农业人工智能 多模态大模型 智慧农业 精准农业","农业人工智能 多模态大模型","农业人工智能多模态大模型智慧农业精准农业-2165","10.1016\u002Fj.engappai.2026.116113",{"doi":534,"openalex_id":536,"authors":537,"venue":352,"cited_by_count":84,"oa_url":502,"card":557,"direction":89,"ingested_from":91},"W7212180335",[538,540,543,545,548,550,552,554],{"name":539,"orcid":9},"KeJian Yu",{"name":541,"orcid":542},"Wentao Ma","https:\u002F\u002Forcid.org\u002F0000-0002-2781-1693",{"name":544,"orcid":9},"Shichao Jin",{"name":546,"orcid":547},"Lu Liu","https:\u002F\u002Forcid.org\u002F0000-0002-3170-9376",{"name":549,"orcid":9},"Yuwei Wang",{"name":551,"orcid":9},"Weiwei Wang",{"name":553,"orcid":9},"Longzhe Quan",{"name":555,"orcid":556},"Lichuan Gu","https:\u002F\u002Forcid.org\u002F0000-0002-3768-8203",{"tldr":558,"method":559,"finding":560,"direction":89,"opportunity":561},"提出农田遥感图文跨模态检索任务，构建数据集并提出FarmMGR框架实现语言驱动检索。","构建67,806图文对FRS-CMR数据集，用多粒度对齐与记忆增强对比学习。","FarmMGR平均召回率28.57%，优于现有图文检索方法，验证语言驱动农田遥感检索可行。","可探索多模态大模型生成数据的噪声校正、跨区域泛化及细粒度农田语义检索基准。","2026-09-11T23:30:30.005240Z"]