[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-641":3,"related-641":51},{"id":4,"title":5,"url":6,"summary":7,"summary_zh":8,"content":9,"source_name":10,"source_url":8,"published_at":11,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":15,"score_detail":16,"sources":24,"tags":26,"primary_topic":8,"secondary_topics":31,"classification":32,"search_phrases":45,"slug":48,"view_count":49,"doi":8,"paper":8,"created_at":50},641,"SAR2Agri:基于纯 SAR 强度图像的自监督学习管道,作物类型制图 IoU 达 84.9%","https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.11142","arXiv 8-11 上线论文提出首个仅使用 SAR 强度图像的农业应用自监督学习管道(SAR2Agri),作者:Moti Rattan Gupta、Anupam Sobti。通过改进的掩码+课程学习时间前置任务,从 SAR 数据中捕获物候特征。在 SICKLE 基准上,最终模型在作物类型制图上达到 84.9% IoU,优于光学基线 15.3 个百分点、优于现有 SAR 基线 2.2 个百分点,展示了 SAR 强度编码器对农业监测预训练的有效性。[预印本]",null,"[View PDF](https:\u002F\u002Farxiv.org\u002Fpdf\u002F2608.11142)[HTML (experimental)](https:\u002F\u002Farxiv.org\u002Fhtml\u002F2608.11142v1)\n\n> Abstract:Agricultural monitoring faces unique challenges, arising from the landscape's complex temporal, phenological, and climate dynamics, yet monitoring them is critical for ensuring food security. Synthetic Aperture Radar (SAR) satellites offer all-weather day-night imaging capability supporting key monitoring tasks including crop type mapping, yield prediction and phenological event detection. Existing multimodal remote sensing foundation models including TerraMind and CopernicusFM learn SAR representations by grounding them in optical imagery using joint encoding and contrastive learning techniques, while SAR-specific foundation models such as SAR-JEPA, SARMAE, and SAR-W-MixMAE primarily focus on target detection, flood mapping, and land cover classification applications. Recent work has introduced phenology inspired temporal pretext tasks with optical imagery which has shown strong performance on agricultural downstream tasks. In this work, we propose the first self-supervised learning pipeline focused on using only SAR intensity imagery for agricultural applications. We improve the temporal pretext tasks through masking and curriculum learning to enhance the pretraining pipeline's ability to capture phenological features from SAR. On the SICKLE benchmark, our final model achieves 84.9% IoU on crop type mapping, outperforming optical baselines (by 15.3 pt) and existing SAR baselines (by 2.2 pt), demonstrating the effectiveness of our proposed pipeline for pretraining SAR intensity encoders for agricultural monitoring.\n\nSubjects:Computer Vision and Pattern Recognition (cs.CV)\nCite as:[arXiv:2608.11142](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.11142) [cs.CV]\n(or [arXiv:2608.11142v1](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.11142v1) [cs.CV] for this version)\n[https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.11142](https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.11142)\n\narXiv-issued DOI via DataCite\n\n## Submission history\n\nFrom: Moti Rattan Gupta [[view email](https:\u002F\u002Farxiv.org\u002Fshow-email\u002Fdc70a2d4\u002F2608.11142)] \n\n**[v1]** Tue, 11 Aug 2026 17:01:26 UTC (28,979 KB)","arXiv:2608.11142 (cs.CV) 2026-08-11","2026-08-11T09:01:00Z","论文",10,false,77,{"impact":17,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":23},20,22,18,12,5,1,"提出首个纯SAR强度图像自监督学习管道，作物制图性能显著提升，对农业遥感监测有重要价值。",[25],{"name":10,"url":6},[27,28,29,30],"深度学习","自监督学习","SAR遥感","农业监测",[],{"status":33,"locked":14,"version":34,"method":35,"evidence":36},"unclassified","2026-10-22.2","rules",{"candidates":37,"agricultural_evidence":43,"reason":44},[38],{"code":39,"score":22,"supported":14,"title":40,"abstract":41,"tags":42},"I2",[],[],[27],true,"缺少足够的主题证据",[46,47],"自监督学习 农业监测 深度学习 SAR遥感","自监督学习 农业监测","自监督学习农业监测深度学习SAR遥感-641",3,"2026-08-16T23:55:42.393184Z",{"total":52,"page":22,"page_size":52,"items":53},6,[54,119,164,228,294,347],{"id":55,"title":56,"url":57,"summary":58,"summary_zh":59,"content":8,"source_name":60,"source_url":57,"published_at":61,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":62,"score_detail":63,"sources":68,"tags":70,"primary_topic":75,"secondary_topics":80,"classification":81,"search_phrases":98,"slug":101,"view_count":102,"doi":103,"paper":104,"created_at":118},5585,"Contrastive Self-Supervised Pretraining for Artificial Intelligence Agriculture Weed Detection Under Label Scarcity: A Computer Transfer-Robustness Evaluation","https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.23543814","Weed detection models for precision agriculture are typically trained on labelled field imagery that is expensive to obtain and geographically narrow, so that a model fitted in one region may fail silently when deployed in another. Contrastive self-supervised pretraining has been proposed as a remedy because it learns representations from unlabelled images, but the conditions under which the resulting representations transfer robustly across domains remain poorly characterised. We define a representation-level robustness functional that separates invariance to nuisance transformations from sensitivity to class-discriminative structure, and we derive a bound on target-domain error in terms of source-domain error, representation stability and a domain-divergence term. The bound implies that contrastive pretraining helps only when the augmentation distribution used during pretraining overlaps with the transformation set that distinguishes source from target domains; otherwise the learned invariances suppress exactly the features the downstream task requires. We operationalise the framework as a transfer-robustness evaluation protocol with three components: a label-budget sweep, a domain-shift ladder and a representation-stability diagnostic. The protocol is specified as a design template rather than executed on new field data, and we report the analytical operating characteristics of the evaluation under a declared variance structure. The main contribution is a precise statement of when contrastive pretraining is expected to improve weed-detection transfer and when it is expected to fail, together with an evaluation design that makes that expectation testable.","精准农业中的杂草检测模型通常基于标注田间影像训练，而这类影像获取成本高且地理覆盖范围狭窄，因此在一个地区拟合的模型部署到另一地区时可能悄然失效。对比自监督预训练被认为是一种补救方案，因为它能够从无标注图像中学习表征，但所得表征在何种条件下能够跨域稳健迁移，目前仍缺乏充分刻画。我们定义了一个表征层面的鲁棒性泛函，将对外扰变换的不变性与对类别判别结构的敏感性区分开来，并推导了目标域误差的上界，该上界由源域误差、表征稳定性和一个域散度项构成。该上界表明，对比预训练只有在预训练所用增强分布与区分源域和目标域的变换集合存在重叠时才有帮助；否则，所学到的不变性恰好会抑制下游任务所需的特征。我们将该框架具体化为一个迁移鲁棒性评估协议，包含三个组成部分：标签预算扫描、域偏移阶梯和表征稳定性诊断。该协议被规定为一种设计模板，而非在新的田间数据上执行，我们在声明的方差结构下报告了该评估的解析运行特性。主要贡献在于精确陈述了对比预训练何时有望改善杂草检测迁移、何时预期会失败，并给出了一种使该预期可检验的评估设计。","Zenodo (CERN European Organization for Nuclear Research)","2026-10-21T00:00:00Z",62,{"impact":64,"substance":19,"depth":65,"authority":66,"freshness":13,"relevant":22,"comment":67},8,17,9,"提出对比自监督预训练在标签稀缺下跨域迁移的可检验评估框架，方法学有新意，但仅为设计模板、未在真实田间数据上验证，属细分领域理论进展。",[69],{"name":60,"url":57},[71,72,28,73,74],"智慧农业","农业人工智能","杂草检测","迁移学习",{"code":39,"slug":76,"name":77,"path":78,"parent_code":79},"agricultural-ai","农业AI与大模型","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-ai","I",[],{"status":82,"locked":14,"version":34,"method":35,"evidence":83},"classified",{"candidates":84,"agricultural_evidence":43},[85,90],{"code":39,"score":13,"supported":43,"title":86,"abstract":88,"tags":89},[87],"artificial intelligence",[],[72,74],{"code":91,"score":92,"supported":43,"title":93,"abstract":94,"tags":97},"I1",7,[],[95,96],"精准农业","precision agriculture",[71],[99,100],"对比自监督 杂草检测","杂草检测 迁移鲁棒性","对比自监督杂草检测-5585",0,"10.5281\u002Fzenodo.23543814",{"doi":103,"openalex_id":105,"authors":106,"venue":60,"cited_by_count":102,"oa_url":57,"card":111,"direction":115,"ingested_from":117},"W7223090735",[107,109],{"name":108,"orcid":8},"Abebe Gebremedhin Desta",{"name":110,"orcid":8},"Kebede Mekonnen Tadesse",{"tldr":112,"method":113,"finding":114,"direction":115,"opportunity":116},"提出对比自监督预训练在杂草检测跨域迁移中何时有效的理论框架与评估协议。","定义表示级鲁棒性泛函，推导目标域误差上界，设计含标签预算扫描与域偏移阶梯的评估协","仅当预训练增强分布与源-目标域差异变换集重叠时，对比预训练才有助于迁移，否则会抑制任务所需特征。","农业人工智能与决策模型","可将该理论框架落地为真实田间多区域杂草检测实验，验证增强分布重叠假设并量化迁移失败边界。","openalex","2026-10-22T11:21:00.352502Z",{"id":120,"title":121,"url":122,"summary":123,"summary_zh":8,"content":8,"source_name":124,"source_url":8,"published_at":125,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":15,"score_detail":126,"sources":129,"tags":131,"primary_topic":8,"secondary_topics":134,"classification":135,"search_phrases":151,"slug":154,"view_count":102,"doi":8,"paper":155,"created_at":163},5531,"A100：面向农业植物图像分析的混合视觉基础模型","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.22991","研究者在覆盖100余种作物、300万张无标注田间图像上自监督预训练A100，采用CNN与Transformer混合骨干融合局部与全局特征。仅用1%标注数据（约每类284张）作物分类准确率达98.01%，在困难数据集PlantDoc上优于全部基线，并在1%标注条件下比公开农业基础模型高14个百分点以上。预印本，未经同行评审。","arXiv 预印本 2026-10-16","2026-10-16T00:00:00Z",{"impact":18,"substance":127,"depth":19,"authority":64,"freshness":52,"relevant":22,"comment":128},23,"大规模田间图像自监督预训练，少样本分类表现突出，方法新颖且数据规模可观，但属未经同行评审的预印本，权威性有限。",[130],{"name":124,"url":122},[71,72,28,132,133],"作物分类","视觉基础模型",[],{"status":33,"locked":14,"version":34,"method":35,"evidence":136},{"candidates":137,"agricultural_evidence":43,"reason":44},[138,142,146],{"code":91,"score":22,"supported":14,"title":139,"abstract":140,"tags":141},[],[],[71],{"code":39,"score":22,"supported":14,"title":143,"abstract":144,"tags":145},[],[],[72],{"code":147,"score":22,"supported":14,"title":148,"abstract":149,"tags":150},"I3",[],[],[132],[152,153],"A100 农业视觉基础模型","PlantDoc 作物分类","A100农业视觉基础模型-5531",{"doi":8,"openalex_id":8,"authors":156,"venue":8,"cited_by_count":102,"oa_url":8,"card":157,"direction":115,"ingested_from":162},[],{"tldr":158,"method":159,"finding":160,"direction":115,"opportunity":161},"构建混合视觉基础模型A100，用百万级田间图像自监督预训练实现少标注作物图像分析。","300万张田间图像自监督预训练，CNN与Transformer混合骨干融合局部与","仅用1%标注数据作物分类准确率达98.01%，困难数据集上优于全部基线。","可探索该基础模型向病虫害检测、表型测量等下游任务的少样本迁移与轻量化部署。","agent","2026-10-22T00:07:09.646474Z",{"id":165,"title":166,"url":167,"summary":168,"summary_zh":169,"content":8,"source_name":170,"source_url":167,"published_at":171,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":102,"score_detail":172,"sources":174,"tags":176,"primary_topic":8,"secondary_topics":179,"classification":180,"search_phrases":202,"slug":205,"view_count":102,"doi":206,"paper":207,"created_at":227},5488,"Application of radar remote sensing for cyclone damage mapping in Bangladesh: a coherence-based approach","https:\u002F\u002Fdoi.org\u002F10.5194\u002Fnhess-26-5103-2026","Abstract. Rapid and reliable assessment of cyclone-induced damage remains a major challenge in coastal regions where cloud cover and accessibility limit conventional approaches. This study presents a coherence-based Synthetic Aperture Radar (SAR) framework for identifying as well as quantifying cyclone-induced surface changes in coastal districts in Bangladesh. Multi-temporal Sentinel-1 data were analyzed using interferometric coherence to capture spatial and temporal variations associated with five major cyclones: Amphan (2020), Sitrang (2022), Mocha (2023), Hamoon (2023), and Remal (2024). A time-series approach incorporating multiple pre-, during-, and post-event image pairs was implemented to distinguish cyclone-driven changes from natural surface variability and to improve detection reliability. The results reveal a consistent decline in coherence during cyclone events followed by gradual recovery. Spatial analysis indicates that areas of significant coherence loss correspond to zones affected by flooding, vegetation damage, and structural disruption. Estimated damage extents varied across cyclones, with Remal (2024) showing the largest affected area (~601 km²), followed by Mocha (~474 km²), Amphan (~288 km²), Sitrang (~122 km²), and Hamoon (~117 km²). Validation using field observations and secondary data yielded an overall classification accuracy of 94 %, confirming the robustness of the coherence-based approach. Complementary inundation mapping further demonstrated that coherence captures a broader spectrum of cyclone impacts beyond flood extent alone, including wind-induced damage and geomorphic changes. This study establishes multi-temporal SAR coherence analysis as a scalable, cloud-independent, and operationally effective tool for rapid post-cyclone damage assessment. The proposed framework enhances the reliability of disaster monitoring in complex coastal environments and offers strong potential for integration into near-real-time disaster response systems. Future research should focus on integrating multi-sensor datasets and long-term coherence time series to further refine damage characterization and reduce uncertainty.","摘要：在云覆盖和可达性限制传统方法的沿海地区，快速可靠地评估气旋所致损害仍是一项重大挑战。本研究提出了一种基于相干性的合成孔径雷达（SAR）框架，用于识别并量化孟加拉国沿海地区气旋引起的地表变化。利用干涉相干性对多时相Sentinel-1数据进行分析，以捕捉与五次主要气旋相关的时间和空间变化：安潘（Amphan，2020年）、西特朗（Sitrang，2022年）、穆查（Mocha，2023年）、哈蒙（Hamoon，2023年）和雷马尔（Remal，2024年）。研究采用了一种整合多个事件前、事件期间和事件后影像对的时间序列方法，以将气旋驱动变化与自然地表的变异性区分开来，并提高检测可靠性。结果表明，气旋事件期间相干性持续下降，随后逐渐恢复。空间分析表明，相干性显著丧失的区域与受洪水、植被破坏和结构损毁影响的区域相对应。各气旋的估计受损范围有所不同，其中雷马尔（2024年）受影响面积最大（约601 km²），其次为穆查（约474 km²）、安潘（约288 km²）、西特朗（约122 km²）和哈蒙（约117 km²）。利用实地观测和二手数据进行的验证获得了94%的总体分类精度，证实了基于相干性方法的稳健性。补充的淹没制图进一步表明，相干性捕捉到了除洪水范围之外的更广泛的气旋影响，包括风致损害和地貌变化。本研究确立了多时相SAR相干性分析作为一种可扩展、不依赖云覆盖且业务上有效的工具，用于气旋后快速损害评估。所提出的框架提高了复杂沿海环境中灾害监测的可靠性，并具有将其集成到近实时灾害响应系统中的强大潜力。未来研究应侧重于整合多传感器数据集和长期相干性时间序列，以进一步完善损害表征并减少不确定性。","Natural hazards and earth system sciences","2026-10-19T00:00:00Z",{"impact":102,"substance":102,"depth":102,"authority":102,"freshness":102,"relevant":102,"comment":173},"该研究聚焦孟加拉国海岸带气旋灾害的SAR相干性制图，虽涉及遥感技术，但与三农、农业信息化、数字乡村等主题无直接关联，不宜进入每日精选。",[175],{"name":170,"url":167},[29,177,178],"遥感监测","灾害评估",[],{"status":181,"locked":14,"version":34,"method":35,"evidence":182},"needs_review",{"candidates":183,"agricultural_evidence":14,"reason":201},[184,189,196],{"code":147,"score":66,"supported":43,"title":185,"abstract":187,"tags":188},[186],"remote sensing",[],[177],{"code":190,"score":52,"supported":43,"title":191,"abstract":192,"tags":195},"I4",[],[193,194],"空间分析","spatial analysis",[],{"code":197,"score":22,"supported":14,"title":198,"abstract":199,"tags":200},"H7",[],[],[178],"农业相关性需要复核",[203,204],"Sentinel-1 气旋 灾害制图","雷达遥感 孟加拉湾 气旋","Sentinel-1气旋灾害制图-5488","10.5194\u002Fnhess-26-5103-2026",{"doi":206,"openalex_id":208,"authors":209,"venue":170,"cited_by_count":102,"oa_url":220,"card":221,"direction":225,"ingested_from":117},"W7160271999",[210,213,216,218],{"name":211,"orcid":212},"Shadmin Islam Prova","https:\u002F\u002Forcid.org\u002F0000-0001-8549-1758",{"name":214,"orcid":215},"Md. Shahoriar Sarker","https:\u002F\u002Forcid.org\u002F0009-0006-5131-1214",{"name":217,"orcid":8},"Rosdania Islam",{"name":219,"orcid":8},"Ridad Ibne Mahfuz","https:\u002F\u002Fnhess.copernicus.org\u002Farticles\u002F26\u002F5103\u002F2026\u002Fnhess-26-5103-2026.pdf",{"tldr":222,"method":223,"finding":224,"direction":225,"opportunity":226},"用Sentinel-1多时相干涉相干性识别并量化孟加拉湾五次气旋的地表破坏。","Sentinel-1多时相InSAR相干性时间序列，结合实地与二手数据验证。","气旋期间相干性显著下降后逐步恢复，分类精度94%，Remal受灾面积最大约601km²。","农业遥感与作物表型","可将相干性框架迁移至农田风灾与洪涝受损制图，融合多源SAR与长时序相干性降低不确定性。","2026-10-21T23:30:23.509482Z",{"id":229,"title":230,"url":231,"summary":232,"summary_zh":233,"content":8,"source_name":234,"source_url":231,"published_at":235,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":236,"score_detail":237,"sources":240,"tags":242,"primary_topic":245,"secondary_topics":251,"classification":253,"search_phrases":273,"slug":276,"view_count":102,"doi":277,"paper":278,"created_at":293},5387,"LIGHTWEIGHT FEATURE FUSED ATTENTION DRIVEN MULTI LEVEL HYBRID MOBILENET V2 FRAMEWORK FOR EARLY TOMATO LEAF DISEASE DETECTION","https:\u002F\u002Fdoi.org\u002F10.62441\u002Factainnovations.vi.1918","It is crucial to find and treat tomato crop diseases early and accurately to improve crop efficiency and make sure there is no scarcity of crop. However, computerized analysis is provocative due to lighting differences, compound backgrounds, and graphic likenesses among infection regions. Research work proposed a lightweight Attention Driven Multi Level Hybrid MobileNet-V2 algorithm that fuses the handcrafted texture structures with Deep Learning (DL) features, then image preprocessing is incorporated using Gaussian, Bilateral, and Guided filters to minimize noise by preserving the disease lesions. LAB color model along with Contrast Limited Adaptive Histogram Equalization (CLAHE) are used to improve the contrast. Otsu thresholding is involved to effectively separate the leaf segment from diseased parts. Local Binary Pattern (LBP) and Gray Level Co-occurrence Matrix (GLCM) descriptors are used to obtain fine-grained features, while high level features are gained by pretrained MobileNet-V2 model. Features fusion is fed to the fully connected layers of deep learning model for training and validation purpose. Results have revealed that the suggested model attains better classification performance as compared to existing models, making it an effective explanation for real time tomato leaf disease detection for smart agriculture applications.","早期准确发现并治疗番茄作物病害，对于提高作物效率、确保作物不短缺至关重要。然而，由于光照差异、复杂背景以及感染区域之间的图形相似性，计算机化分析具有挑战性。研究提出了一种轻量级注意力驱动多层混合MobileNet-V2算法，该算法将手工纹理结构与深度学习（DL）特征相融合，然后结合高斯、双边和引导滤波器进行图像预处理，以在保留病害病变的同时最小化噪声。采用LAB颜色模型和对比度受限自适应直方图均衡化（CLAHE）来提高对比度。引入Otsu阈值分割以有效分离叶片片段与病害部分。使用局部二值模式（LBP）和灰度共生矩阵（GLCM）描述符获取细粒度特征，同时通过预训练的MobileNet-V2模型获取高层特征。特征融合后输入深度学习模型的全连接层进行训练和验证。结果表明，与现有模型相比，所提出的模型取得了更好的分类性能，使其成为智慧农业应用中实时番茄叶片病害检测的有效解决方案。","Acta Innovations","2026-10-18T00:00:00Z",64,{"impact":20,"substance":19,"depth":238,"authority":13,"freshness":64,"relevant":22,"comment":239},16,"提出融合手工纹理特征与MobileNet-V2的轻量番茄叶病检测方法，有一定方法新颖性，但属单篇论文、应用范围有限，未达每日精选门槛。",[241],{"name":234,"url":231},[71,72,27,243,244],"番茄病害","图像识别",{"code":246,"slug":247,"name":248,"path":249,"parent_code":250},"B1","vegetables","蔬菜","\u002Ftopic\u002Fhorticulture\u002Fvegetables","B",[252],{"code":39,"slug":76,"name":77,"path":78,"parent_code":79},{"status":82,"locked":14,"version":34,"method":35,"evidence":254},{"candidates":255,"agricultural_evidence":43},[256,263,268],{"code":246,"score":257,"supported":43,"title":258,"abstract":260,"tags":262},15,[259],"tomato",[261,259],"番茄",[243],{"code":39,"score":66,"supported":43,"title":264,"abstract":265,"tags":267},[],[27,266],"deep learning",[72,244,27],{"code":91,"score":92,"supported":43,"title":269,"abstract":270,"tags":272},[],[71,271],"smart agriculture",[71],[274,275],"MobileNet-V2 番茄叶片病害","番茄叶病 深度学习 检测","MobileNet-V2番茄叶片病害-5387","10.62441\u002Factainnovations.vi.1918",{"doi":277,"openalex_id":279,"authors":280,"venue":234,"cited_by_count":102,"oa_url":286,"card":287,"direction":292,"ingested_from":117},"W7222977034",[281,283],{"name":282,"orcid":8},"Mrs. Vijetalaxmi A Hegde",{"name":284,"orcid":285},"Abhay Ramakant Kasetwar","https:\u002F\u002Forcid.org\u002F0000-0002-5800-8764","https:\u002F\u002Factainnovations.com\u002Findex.php\u002Fpub\u002Farticle\u002Fdownload\u002F1918\u002F981",{"tldr":288,"method":289,"finding":290,"direction":115,"opportunity":291},"提出轻量注意力多级混合MobileNet-V2框架，融合手工纹理与深度学习特征实现番茄叶片病害早期检","高斯\u002F双边\u002F引导滤波去噪，LAB+CLAHE增强，Otsu分割，LBP与GLCM","所提模型分类性能优于现有模型，可有效用于智慧农业实时番茄叶片病害检测。","可探索多作物跨域泛化、田间复杂光照下轻量模型部署与可解释性，及病害早期分级检测。","智慧农业 \u002F 农业物联网","2026-10-20T23:30:10.424511Z",{"id":295,"title":296,"url":297,"summary":298,"summary_zh":299,"content":8,"source_name":300,"source_url":297,"published_at":235,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":301,"score_detail":302,"sources":304,"tags":306,"primary_topic":309,"secondary_topics":310,"classification":311,"search_phrases":333,"slug":336,"view_count":102,"doi":337,"paper":338,"created_at":346},5385,"A Comprehensive Review of Deep Learning and Computer Vision Techniques for Fruit Quality Assessment and Classification","https:\u002F\u002Fdoi.org\u002F10.25258\u002Fijddt.16.75s.151","Fruit image analysis has become an essential research area in precision agriculture, enabling automated fruit classification, quality grading, maturity estimation, disease diagnosis, yield prediction, and intelligent harvesting through advances in computer vision and artificial intelligence (AI).Recent progress in imaging technologies, large-scale benchmark datasets, and deep learning has substantially improved the accuracy, robustness, and scalability of automated fruit analysis systems.This survey presents a comprehensive and systematic review of image processing and AI techniques for fruit image analysis.It covers the complete image analysis pipeline, including image acquisition, preprocessing, enhancement, color space transformation, segmentation, feature extraction, and handcrafted texture descriptors.Furthermore, conventional machine learning algorithms are reviewed alongside state-of-the-art deep learning architectures, including Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), and Vision Mamba models, for applications such as fruit recognition, quality assessment, disease detection, ripeness estimation, and object detection.Publicly available benchmark datasets, evaluation metrics, and performance comparison strategies are also summarized to facilitate objective assessment of existing approaches.In addition, the survey discusses key research challenges, including illumination variations, occlusion, complex orchard environments, limited annotated datasets, computational complexity, and domain adaptation, while highlighting emerging research trends such as self-supervised learning, multimodal learning, explainable AI, lightweight deep networks, foundation models, and edge intelligence for real-time deployment.By providing a unified review of methodologies, datasets, evaluation protocols, challenges, and future research directions, this survey offers a comprehensive reference for researchers and practitioners developing intelligent and scalable fruit image analysis systems for precision agriculture.","水果图像分析已成为精准农业中的一个重要研究领域，借助计算机视觉和人工智能（AI）的进步，实现了果实自动分类、品质分级、成熟度估计、病害诊断、产量预测和智能采收。成像技术、大规模基准数据集和深度学习的最新进展显著提升了自动化水果分析系统的准确性、鲁棒性和可扩展性。本综述对用于水果图像分析的图像处理和AI技术进行了全面而系统的回顾。内容涵盖完整的图像分析流程，包括图像采集、预处理、增强、颜色空间变换、分割、特征提取和手工纹理描述符。此外，综述了传统机器学习算法以及最先进的深度学习架构，包括卷积神经网络（CNN）、视觉Transformer（ViT）和视觉Mamba模型，用于果实识别、品质评估、病害检测、成熟度估计和目标检测等应用。同时总结了公开可用的基准数据集、评价指标和性能比较策略，以便对现有方法进行客观评估。此外，本综述讨论了关键研究挑战，包括光照变化、遮挡、复杂果园环境、标注数据集有限、计算复杂度和域适应，并重点介绍了新兴研究趋势，如自监督学习、多模态学习、可解释AI、轻量级深度网络、基础模型和面向实时部署的边缘智能。通过提供方法、数据集、评价协议、挑战和未来研究方向的统一综述，本综述为开发面向精准农业的智能且可扩展的水果图像分析系统的研究人员和实践者提供了全面的参考。","International Journal of Drug Delivery Technology",56,{"impact":64,"substance":19,"depth":238,"authority":52,"freshness":64,"relevant":22,"comment":303},"系统性综述覆盖CNN、ViT、Vision Mamba等前沿方法，信息量足，但期刊权威性一般且偏学术综述，公共价值有限，适合主题聚合而非每日精选头条。",[305],{"name":300,"url":297},[71,72,307,27,308],"计算机视觉","水果品质检测",{"code":39,"slug":76,"name":77,"path":78,"parent_code":79},[],{"status":82,"locked":14,"version":34,"method":35,"evidence":312},{"candidates":313,"agricultural_evidence":43},[314,323,327],{"code":39,"score":315,"supported":43,"title":316,"abstract":318,"tags":322},28,[266,317],"computer vision",[319,27,320,307,321,266,87,317],"目标检测","机器学习","machine learning",[72,27,307],{"code":91,"score":92,"supported":43,"title":324,"abstract":325,"tags":326},[],[95,96],[71],{"code":328,"score":49,"supported":43,"title":329,"abstract":330,"tags":332},"B2",[],[331],"orchard",[],[334,335],"水果品质 计算机视觉 深度学习","农业人工智能 水果品质检测 计算机视觉 智慧农业","水果品质计算机视觉深度学习-5385","10.25258\u002Fijddt.16.75s.151",{"doi":337,"openalex_id":339,"authors":340,"venue":300,"cited_by_count":102,"oa_url":297,"card":341,"direction":292,"ingested_from":117},"W7222892831",[],{"tldr":342,"method":343,"finding":344,"direction":225,"opportunity":345},"系统综述深度学习与计算机视觉在水果品质评估与分级中的全流程方法。","综述图像采集到分割、特征提取及CNN、ViT、Vision Mamba等模型与公","深度学习显著提升水果分析精度与鲁棒性，但光照、遮挡和小样本仍是主要瓶颈。","可探索自监督与轻量化模型在复杂果园遮挡、小样本场景下的边缘实时部署。","2026-10-20T23:30:10.066042Z",{"id":348,"title":349,"url":350,"summary":351,"summary_zh":352,"content":8,"source_name":300,"source_url":350,"published_at":353,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":102,"score_detail":354,"sources":356,"tags":358,"primary_topic":8,"secondary_topics":359,"classification":360,"search_phrases":367,"slug":370,"view_count":102,"doi":371,"paper":372,"created_at":384},5242,"Enhanced Convolutional Neural Network Architecture with Attention-Guided Feature Refinement for Robust Image Classification","https:\u002F\u002Fdoi.org\u002F10.25258\u002Fijddt.16.88s.43","Image classification remains a foundational problem in computer vision, underpinning applications that range from medical diagnostics to autonomous navigation. Although deep convolutional neural networks (CNNs) have substantially advanced classification accuracy over the last decade, standard architectures such as VGG, ResNet, and plain CNNs often struggle to jointly optimize accuracy, parameter efficiency, and robustness to intra-class variation. This paper proposes an AttentionGuided Refinement Network (AGR-Net), a CNN architecture that integrates a lightweight channel-and-spatial attention module with residual feature refinement blocks to improve discriminative feature learning while maintaining a compact parameter footprint. The proposed model is evaluated on the CIFAR-10 benchmark dataset using a rigorous experimental protocol that includes data augmentation, learning-rate scheduling, and five-fold cross-validation. Comparative experiments against baseline architectures, including a plain CNN, VGG-16, ResNet-18, and MobileNetV2, demonstrate that AGR-Net achieves a top-1 accuracy of 94.6%, outperforming the strongest baseline by 1.8 percentage points while using 27% fewer parameters than ResNet-18. Ablation studies confirm that the attention module and refinement blocks each contribute measurably to the accuracy gain. The results indicate that attention-guided refinement offers a favorable trade-off between accuracy, efficiency, and robustness, making the proposed architecture suitable for deployment in resource-constrained and real-time image classification systems. The paper further discusses the advantages, practical applications, limitations, and future research directions of the proposed approach.","图像分类仍然是计算机视觉中的一个基础性问题，支撑着从医学诊断到自主导航等诸多应用。尽管深度卷积神经网络（CNN）在过去十年中大幅提升了分类准确率，但VGG、ResNet和普通CNN等标准架构往往难以在准确率、参数效率和对类内变化的鲁棒性之间实现联合优化。本文提出了一种注意力引导细化网络（AttentionGuided Refinement Network，AGR-Net），这是一种CNN架构，它将轻量级的通道与空间注意力模块与残差特征细化模块相结合，以改进判别性特征学习，同时保持紧凑的参数规模。所提出的模型在CIFAR-10基准数据集上使用严格的实验方案进行评估，该方案包括数据增强、学习率调度和五折交叉验证。与基线架构（包括普通CNN、VGG-16、ResNet-18和MobileNetV2）的对比实验表明，AGR-Net实现了94.6%的top-1准确率，比最强基线高出1.8个百分点，同时使用的参数比ResNet-18少27%。消融研究证实，注意力模块和细化模块各自对准确率提升都有可测量的贡献。结果表明，注意力引导细化在准确率、效率和鲁棒性之间提供了有利的权衡，使所提出的架构适合部署在资源受限和实时的图像分类系统中。本文进一步讨论了所提出方法的优势、实际应用、局限性和未来研究方向。","2026-10-17T00:00:00Z",{"impact":102,"substance":102,"depth":102,"authority":102,"freshness":102,"relevant":102,"comment":355},"该文为通用计算机视觉论文，未涉及农业场景，与三农及农业信息化无关，不宜入选。",[357],{"name":300,"url":350},[72,27,244],[],{"status":181,"locked":14,"version":34,"method":35,"evidence":361},{"candidates":362,"agricultural_evidence":14,"reason":201},[363],{"code":39,"score":66,"supported":43,"title":364,"abstract":365,"tags":366},[],[307,317],[72,244,27],[368,369],"农业人工智能 图像识别 深度学习","农业人工智能 图像识别","农业人工智能图像识别深度学习-5242","10.25258\u002Fijddt.16.88s.43",{"doi":371,"openalex_id":373,"authors":374,"venue":300,"cited_by_count":102,"oa_url":350,"card":379,"direction":292,"ingested_from":117},"W7222796569",[375,377],{"name":376,"orcid":8},"KRISHNA SANDEEP JONNALA",{"name":378,"orcid":8},"K VENKAT RAJU",{"tldr":380,"method":381,"finding":382,"direction":115,"opportunity":383},"提出注意力引导细化网络AGR-Net，在CIFAR-10上兼顾精度与参数效率。","轻量通道-空间注意力模块+残差特征细化块，五折交叉验证对比基线。","AGR-Net达94.6%准确率，超最强基线1.8个百分点且参数比ResNet-18少27%。","可将该轻量注意力架构迁移至作物病害识别等农业图像任务，验证田间复杂场景鲁棒性。","2026-10-18T23:30:13.674128Z"]