[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-2868":3,"related-2868":90},{"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":28,"primary_topic":9,"secondary_topics":34,"classification":35,"search_phrases":58,"slug":61,"view_count":22,"doi":62,"paper":63,"created_at":89},2868,"MCLC-NET: Multimodal Continual Learning for Leaf Counting","https:\u002F\u002Fdoi.org\u002F10.48550\u002Farxiv.2609.18129","Leaf counting is an important task in plant phenotyping for monitoring plant growth and estimating crop yield. Most existing methods rely on RGB images, but their performance is often affected by occlusion, lighting variations, and other real-world challenges. Additional modalities, such as depth and thermal images, can provide useful complementary information. However, multimodal leaf counting remains underexplored. Also, many existing methods assume that all training data are available simultaneously, which is impractical in real agricultural settings, where data is collected over time from multiple sources. To address these challenges, we propose MCLC-NET, a multimodal continual learning framework for leaf counting. It learns tasks sequentially using a memory-based strategy with a memory buffer to retain important samples from previous tasks. We also introduce MMLC, a real-world multimodal leaf-counting dataset designed for a domain incremental scenario (DIS) in CL. It contains RGB, depth, and thermal images collected across different crop types under varying environmental conditions, arranged in three orderings: crop-wise, time-wise, and mixed. Experimental results, averaged over three random seeds, demonstrate that MCLC-NET consistently outperforms existing methods across all three task orderings, achieving the lowest AMSE of 0.675$\\pm$0.027, 0.542$\\pm$0.069, and 0.745$\\pm$0.057, respectively.","叶片计数是植物表型分析中的一项重要任务，用于监测植物生长和估算作物产量。现有方法大多依赖RGB图像，但其性能往往受到遮挡、光照变化及其他现实挑战的影响。深度图像和热成像等其他模态可以提供有用的互补信息。然而，多模态叶片计数仍未被充分探索。此外，许多现有方法假设所有训练数据可同时获取，这在实际农业场景中并不现实，因为数据是随时间从多个来源收集的。为应对这些挑战，我们提出了MCLC-NET，一种用于叶片计数的多模态持续学习框架。该框架采用基于记忆的策略，通过记忆缓冲区保留先前任务中的重要样本，从而按顺序学习任务。我们还引入了MMLC，一个面向持续学习中域增量场景（DIS）设计的真实世界多模态叶片计数数据集。该数据集包含在不同环境条件下跨不同作物类型采集的RGB、深度和热成像图像，并按三种顺序排列：按作物、按时间和混合。在三个随机种子上的平均实验结果表明，MCLC-NET在所有三种任务顺序下均持续优于现有方法，分别取得了最低的AMSE，为0.675±0.027、0.542±0.069和0.745±0.057。",null,"arXiv (Cornell University)","2026-09-16T00:00:00Z","论文",25,false,74,{"impact":17,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":23},12,22,18,13,9,1,"提出多模态持续学习叶片计数框架并发布真实农业数据集，方法新颖、实验扎实，但属细分领域学术进展，产业影响有限。",[25,26],{"name":10,"url":6},{"name":10,"url":27},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.18129",[29,30,31,32,33],"智慧农业","农业人工智能","作物表型","多模态学习","叶片计数",[],{"status":36,"locked":14,"version":37,"method":38,"evidence":39},"unclassified","2026-10-22.2","rules",{"candidates":40,"agricultural_evidence":56,"reason":57},[41,46,51],{"code":42,"score":22,"supported":14,"title":43,"abstract":44,"tags":45},"I1",[],[],[29],{"code":47,"score":22,"supported":14,"title":48,"abstract":49,"tags":50},"I2",[],[],[30],{"code":52,"score":22,"supported":14,"title":53,"abstract":54,"tags":55},"I3",[],[],[31],true,"缺少足够的主题证据",[59,60],"MCLC-NET 叶片计数","多模态 持续学习 作物表型","MCLC-NET叶片计数-2868","10.48550\u002Farxiv.2609.18129",{"doi":62,"openalex_id":64,"authors":65,"venue":10,"cited_by_count":81,"oa_url":6,"card":82,"direction":86,"ingested_from":88},"W7213503285",[66,69,71,74,76,79],{"name":67,"orcid":68},"Ruchi Bhatt","https:\u002F\u002Forcid.org\u002F0009-0003-8222-5640",{"name":70,"orcid":9},"Pratibha Kumari",{"name":72,"orcid":73},"Shreya Bansal","https:\u002F\u002Forcid.org\u002F0000-0002-3135-291X",{"name":75,"orcid":9},"Vedant Agnihotri",{"name":77,"orcid":78},"Dwarikanath Mahapatra","https:\u002F\u002Forcid.org\u002F0000-0001-9749-7858",{"name":80,"orcid":9},"Mukesh Saini",0,{"tldr":83,"method":84,"finding":85,"direction":86,"opportunity":87},"提出多模态持续学习框架MCLC-NET用于叶片计数，并发布真实多模态数据集MMLC。","基于记忆缓冲的持续学习策略，融合RGB、深度和热成像三种模态数据。","在三种任务顺序下均优于现有方法，最低AMSE达0.542±0.069。","农业遥感与作物表型","多模态持续学习在农业表型中尚属空白，可探索更多模态融合与动态环境适应策略。","openalex","2026-09-18T23:30:22.580822Z",{"total":91,"page":22,"page_size":91,"items":92},6,[93,196,261,318,408,499],{"id":94,"title":95,"url":96,"summary":97,"summary_zh":98,"content":9,"source_name":99,"source_url":96,"published_at":100,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":101,"score_detail":102,"sources":106,"tags":111,"primary_topic":114,"secondary_topics":119,"classification":120,"search_phrases":168,"slug":171,"view_count":81,"doi":172,"paper":173,"created_at":195},4814,"Wearable Sensors for Smart Agriculture: From Sensing Mechanisms to Crop Phenotyping and Harvesting Applications","https:\u002F\u002Fdoi.org\u002F10.3390\u002Fagriculture16202210","Global food security faces severe pressures from population growth, climate change and resource shortages. Traditional agricultural monitoring relies on manual sampling and offline laboratory testing, which involve considerable time delays and cannot realize continuous, in situ dynamic tracking of crop physiological status, field environment and agricultural product quality. Among the various sensing platforms for agriculture, flexible wearable sensors represent an important branch, featuring light weight, conformability, low power consumption and real-time response. It should be noted that rigid-substrate and silicon-based sensors remain indispensable in many agricultural applications, and this review focuses specifically on flexible wearable sensors while acknowledging the broader landscape of agricultural sensing technologies. Through conformal contact with plant surfaces or wearable integration, they can convert physical, chemical and biological signals of agroecosystems into readable electrical\u002Foptical signals via specific sensing transduction mechanisms. This review systematically categorizes wearable sensors for smart agriculture into two main types, optical and electrical, and elaborates their intrinsic sensing principles, summarizes typical application scenarios covering crop growth monitoring, post-harvest harvesting and agricultural product safety detection, and introduces the full-flow signal preprocessing, feature extraction and intelligent analysis technologies matched with wearable sensing data. Finally, the bottlenecks restricting large-scale field deployment of current wearable sensors are summarized, and the development directions of multi-modal integrated wearable sensing, biomimetic nanosensors inspired by natural plant sensing systems, and artificial intelligence (AI)-assisted agricultural monitoring platforms are prospected, aiming to provide a theoretical reference for the design optimization and popularization of agricultural wearable sensing equipment.","全球粮食安全面临人口增长、气候变化和资源短缺的严峻压力。传统农业监测依赖人工采样和离线实验室检测，存在较大的时间延迟，无法实现对作物生理状态、田间环境和农产品品质的连续、原位动态追踪。在各类农业传感平台中，柔性可穿戴传感器是一个重要分支，具有轻量、共形、低功耗和实时响应等特点。需要指出的是，刚性基底和硅基传感器在许多农业应用中仍不可或缺，本综述聚焦于柔性可穿戴传感器，同时承认农业传感技术的更广泛格局。通过与植物表面的共形接触或可穿戴集成，柔性可穿戴传感器能够借助特定的传感转导机制，将农业生态系统的物理、化学和生物信号转化为可读的电信号\u002F光信号。本综述将面向智慧农业的可穿戴传感器系统性地分为光学和电学两大类，阐述其内在传感原理，总结涵盖作物生长监测、采后收获和农产品安全检测的典型应用场景，并介绍与可穿戴传感数据匹配的全流程信号预处理、特征提取和智能分析技术。最后，总结了制约当前可穿戴传感器大规模田间部署的瓶颈，并展望了多模态集成可穿戴传感、受自然植物传感系统启发的仿生纳米传感器以及人工智能（AI）辅助农业监测平台的发展方向，旨在为农业可穿戴传感设备的设计优化与推广应用提供理论参考。","Agriculture","2026-10-11T00:00:00Z",77,{"impact":19,"substance":103,"depth":19,"authority":20,"freshness":104,"relevant":22,"comment":105},20,8,"系统综述柔性可穿戴传感在智慧农业中的机理与应用，方法梳理扎实，对农业信息化领域有参考价值。",[107,108],{"name":99,"url":96},{"name":109,"url":110},"MDPI Agriculture 16(20) 2210 2026-10-11","https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F20\u002F2210",[29,30,31,112,113],"农产品安全","可穿戴传感器",{"code":52,"slug":115,"name":116,"path":117,"parent_code":118},"remote-sensing","遥感与农情监测","\u002Ftopic\u002Fdigital-agriculture\u002Fremote-sensing","I",[],{"status":121,"locked":14,"version":37,"method":38,"evidence":122},"classified",{"candidates":123,"agricultural_evidence":56},[124,130,135,142,149,155,161],{"code":42,"score":125,"supported":56,"title":126,"abstract":128,"tags":129},15,[127],"smart agriculture",[29,127],[29],{"code":52,"score":21,"supported":56,"title":131,"abstract":133,"tags":134},[132],"crop phenotyping",[],[31],{"code":136,"score":91,"supported":56,"title":137,"abstract":138,"tags":141},"A8",[],[139,140],"粮食安全","food security",[],{"code":143,"score":91,"supported":56,"title":144,"abstract":145,"tags":148},"H7",[],[146,147],"气候变化","climate change",[],{"code":47,"score":150,"supported":56,"title":151,"abstract":152,"tags":154},4,[],[153],"artificial intelligence",[30],{"code":156,"score":150,"supported":56,"title":157,"abstract":158,"tags":160},"I5",[],[159],"农业传感",[113],{"code":162,"score":163,"supported":56,"title":164,"abstract":165,"tags":167},"H6",3,[],[166],"农业生态",[],[169,170],"可穿戴传感器 智慧农业","作物表型 传感器","可穿戴传感器智慧农业-4814","10.3390\u002Fagriculture16202210",{"doi":172,"openalex_id":174,"authors":175,"venue":99,"cited_by_count":81,"oa_url":96,"card":189,"direction":194,"ingested_from":88},"W7221545913",[176,178,181,183,185,187],{"name":177,"orcid":9},"Liuyang Qiu",{"name":179,"orcid":180},"Xindong Wang","https:\u002F\u002Forcid.org\u002F0000-0002-8101-7300",{"name":182,"orcid":9},"Xinxin Wu",{"name":184,"orcid":9},"Jie Yu",{"name":186,"orcid":9},"Hao Li",{"name":188,"orcid":9},"Xinai Zhang",{"tldr":190,"method":191,"finding":192,"direction":86,"opportunity":193},"综述柔性可穿戴传感器在智慧农业中的传感机制、作物表型监测与采收应用。","综述光学与电学两类柔性可穿戴传感器原理、应用及信号智能分析技术。","可穿戴传感器可实现作物生理、环境与品质的连续原位监测，但田间规模化部署仍受限。","多模态集成、仿生纳米传感与AI辅助平台是突破田间规模化监测瓶颈的关键方向。","智慧农业 \u002F 农业物联网","2026-10-13T23:30:12.054798Z",{"id":197,"title":198,"url":199,"summary":200,"summary_zh":201,"content":9,"source_name":202,"source_url":199,"published_at":203,"category":12,"cover_url":9,"hotness":204,"is_selected":14,"score":205,"score_detail":206,"sources":210,"tags":212,"primary_topic":214,"secondary_topics":218,"classification":219,"search_phrases":235,"slug":238,"view_count":163,"doi":239,"paper":240,"created_at":260},3963,"MSO-Det: A multi-level synergistic optimization framework for robust spike object detection in field scenes","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.compag.2026.112473","MSO-Det: A multi-level synergistic optimization framework for robust spike object detection in field scenes。Computers and Electronics in Agriculture","MSO-Det：一种用于田间场景中鲁棒脉冲目标检测的多级协同优化框架。《计算机与电子农业》","Computers and Electronics in Agriculture","2026-10-01T00:00:00Z",10,69,{"impact":17,"substance":19,"depth":207,"authority":208,"freshness":21,"relevant":22,"comment":209},16,14,"发表于农业信息领域核心期刊的穗部目标检测新方法，方法有创新但属细分技术进展，公共影响有限。",[211],{"name":202,"url":199},[29,30,213,31],"目标检测",{"code":47,"slug":215,"name":216,"path":217,"parent_code":118},"agricultural-ai","农业AI与大模型","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-ai",[],{"status":121,"locked":14,"version":37,"method":38,"evidence":220},{"candidates":221,"agricultural_evidence":56},[222,227,231],{"code":47,"score":223,"supported":56,"title":224,"abstract":225,"tags":226},5,[],[213],[30,213],{"code":42,"score":22,"supported":14,"title":228,"abstract":229,"tags":230},[],[],[29],{"code":52,"score":22,"supported":14,"title":232,"abstract":233,"tags":234},[],[],[31],[236,237],"MSO-Det 穗部检测","田间场景 穗目标检测","MSO-Det穗部检测-3963","10.1016\u002Fj.compag.2026.112473",{"doi":239,"openalex_id":241,"authors":242,"venue":202,"cited_by_count":81,"oa_url":9,"card":254,"direction":258,"ingested_from":88},"W7215045657",[243,245,247,249,251],{"name":244,"orcid":9},"Yiqun Wang",{"name":246,"orcid":9},"Shuo Zhang",{"name":248,"orcid":9},"Xiaolong Liu",{"name":250,"orcid":9},"Ze Ji",{"name":252,"orcid":253},"Weikuan Jia","https:\u002F\u002Forcid.org\u002F0000-0001-6242-3269",{"tldr":255,"method":256,"finding":257,"direction":258,"opportunity":259},"提出MSO-Det多级协同优化框架，用于田间场景下鲁棒的穗部目标检测。","多级协同优化框架，结合田间场景图像进行穗目标检测。","该框架提升了田间穗检测的鲁棒性，但摘要未给出具体指标。","农业人工智能与决策模型","可探索多级优化在复杂田间光照与遮挡下的泛化能力及轻量化部署。","2026-10-02T23:30:01.492947Z",{"id":262,"title":263,"url":264,"summary":265,"summary_zh":9,"content":9,"source_name":266,"source_url":264,"published_at":267,"category":12,"cover_url":9,"hotness":204,"is_selected":14,"score":268,"score_detail":269,"sources":271,"tags":273,"primary_topic":9,"secondary_topics":276,"classification":277,"search_phrases":295,"slug":298,"view_count":163,"doi":299,"paper":300,"created_at":317},3528,"Long-term dynamic root monitoring based on classification fusion and attention mechanism","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.biosystemseng.2026.104602","Long-term dynamic root monitoring based on classification fusion and attention mechanism。Biosystems Engineering","Biosystems Engineering","2026-09-25T00:00:00Z",67,{"impact":17,"substance":19,"depth":207,"authority":20,"freshness":104,"relevant":22,"comment":270},"核心期刊论文，方法上有分类融合与注意力机制创新，属农业人工智能细分进展，但应用范围有限，未达重大突破层级。",[272],{"name":266,"url":264},[29,30,274,31,275],"深度学习","根系监测",[],{"status":278,"locked":14,"version":37,"method":38,"evidence":279},"needs_review",{"candidates":280,"agricultural_evidence":14,"reason":294},[281,286,290],{"code":47,"score":282,"supported":14,"title":283,"abstract":284,"tags":285},2,[],[],[30,274],{"code":42,"score":22,"supported":14,"title":287,"abstract":288,"tags":289},[],[],[29],{"code":52,"score":22,"supported":14,"title":291,"abstract":292,"tags":293},[],[],[31],"农业相关性需要复核",[296,297],"Biosystems Engineering 根系监测","动态根系监测 注意力机制","BiosystemsEngineering根系监测-3528","10.1016\u002Fj.biosystemseng.2026.104602",{"doi":299,"openalex_id":301,"authors":302,"venue":266,"cited_by_count":81,"oa_url":9,"card":9,"direction":9,"ingested_from":88},"W7214313105",[303,305,307,309,311,313,315],{"name":304,"orcid":9},"Yinkai Fu",{"name":306,"orcid":9},"Pengchong Zhang",{"name":308,"orcid":9},"Yue Zhao",{"name":310,"orcid":9},"Qiaoling Han",{"name":312,"orcid":9},"Benye Xi",{"name":314,"orcid":9},"Yandong Zhao",{"name":316,"orcid":9},"Jianhui Lin","2026-09-26T23:30:06.143382Z",{"id":319,"title":320,"url":321,"summary":322,"summary_zh":323,"content":9,"source_name":324,"source_url":321,"published_at":325,"category":12,"cover_url":9,"hotness":204,"is_selected":14,"score":326,"score_detail":327,"sources":329,"tags":331,"primary_topic":334,"secondary_topics":339,"classification":340,"search_phrases":359,"slug":362,"view_count":150,"doi":363,"paper":364,"created_at":407},3490,"Balancing accuracy, completeness, and efficiency for rice 3D reconstruction through CBAM-UNet-based multi-view segmentation and camera configuration optimization","https:\u002F\u002Fdoi.org\u002F10.3389\u002Ffpls.2026.1882534","Multi-view 3D reconstruction has been widely applied in plant phenotyping, but the complex canopy structure of rice plants poses significant challenges for reconstruction accuracy, completene7ss, and efficiency. In this study, we developed an optimized workflow for 3D reconstruction of rice using the self-developed Metatlas V1 multi-view imaging platform combined with the COLMAP + OpenMVS pipeline. A CBAM-UNet-based image segmentation model was used to extract plant regions from complex backgrounds, outperforming thresholding, U-Net, and U-Net++, and increasing the number of reconstructed points. Camera configuration optimization was performed by systematically combining equidistant and greedy strategies, resulting in an optimal six-camera setup (#2, #3, #4, #6, #7, and #8, corresponding to −30°, −15°, 0°, +30°, +45°, and +60° relative to the horizontal viewpoint), which provided complementary views of the canopy inner structure and stem base. This configuration reduced the average nearest-neighbor distance to 0.037–0.066 cm across different varieties and growth stages compared with the full 11-camera setup, retained 73–76% of the points, and decreased reconstruction duration by approximately 70–80%. The strong agreement between point-cloud-derived and manually measured plant height and canopy width ( R 2 = 0.989 and 0.946, respectively) supported the accuracy of the point-cloud-derived phenotypic measurements. Overall, integrating image segmentation with camera-configuration optimization offers an accurate and efficient solution for high-throughput 3D phenotyping of rice using Metatlas V1, balancing reconstruction accuracy, point-cloud completeness, and computational efficiency, and may provide a methodological reference for camera-configuration design in other rotational multi-view phenotyping platforms.","多视角三维重建已广泛应用于植物表型分析，但水稻复杂的冠层结构对重建精度、完整性和效率构成了重大挑战。本研究利用自主研发的Metatlas V1多视角成像平台，结合COLMAP + OpenMVS流程，开发了一套优化的水稻三维重建工作流。采用基于CBAM-UNet的图像分割模型从复杂背景中提取植株区域，其性能优于阈值分割、U-Net和U-Net++，并增加了重建点数量。通过系统组合等距策略和贪心策略进行相机配置优化，得到了最优的六相机方案（#2、#3、#4、#6、#7和#8，分别对应相对于水平视角的−30°、−15°、0°、+30°、+45°和+60°），该方案提供了冠层内部结构和茎基部的互补视角。与完整的11相机方案相比，该配置将不同品种和生育期的平均最近邻距离降至0.037–0.066 cm，保留了73–76%的点云，并将重建时间缩短了约70–80%。点云提取的株高和冠幅与人工测量结果高度一致（R²分别为0.989和0.946），验证了点云表型测量的准确性。总体而言，将图像分割与相机配置优化相结合，为利用Metatlas V1进行水稻高通量三维表型分析提供了一种准确高效的解决方案，在重建精度、点云完整性和计算效率之间取得了平衡，并可为其他旋转式多视角表型平台的相机配置设计提供方法学参考。","Frontiers in Plant Science","2026-09-24T00:00:00Z",78,{"impact":207,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":328},"该研究提出结合CBAM-UNet分割与相机配置优化的水稻三维重建流程，方法新颖、数据详实，对高通量作物表型分析有参考价值，但属细分领域进展，时效性高。",[330],{"name":324,"url":321},[29,30,332,31,333],"水稻","三维重建",{"code":335,"slug":336,"name":332,"path":337,"parent_code":338},"A1","rice","\u002Ftopic\u002Ffield-crops\u002Frice","A",[],{"status":121,"locked":14,"version":37,"method":38,"evidence":341},{"candidates":342,"agricultural_evidence":56},[343,347,351,355],{"code":335,"score":125,"supported":56,"title":344,"abstract":345,"tags":346},[336],[332,336],[332],{"code":47,"score":282,"supported":14,"title":348,"abstract":349,"tags":350},[],[],[333,30],{"code":42,"score":22,"supported":14,"title":352,"abstract":353,"tags":354},[],[],[29],{"code":52,"score":22,"supported":14,"title":356,"abstract":357,"tags":358},[],[],[31],[360,361],"Metatlas V1 水稻 三维重建","CBAM-UNet 水稻 图像分割","MetatlasV1水稻三维重建-3490","10.3389\u002Ffpls.2026.1882534",{"doi":363,"openalex_id":365,"authors":366,"venue":324,"cited_by_count":81,"oa_url":321,"card":402,"direction":86,"ingested_from":88},"W7214231705",[367,369,372,374,377,379,381,384,387,390,392,394,397,400],{"name":368,"orcid":9},"Haoyang Zhou",{"name":370,"orcid":371},"Rongjie Chen","https:\u002F\u002Forcid.org\u002F0009-0004-8078-2301",{"name":373,"orcid":9},"Hao Wang",{"name":375,"orcid":376},"Minglu Li","https:\u002F\u002Forcid.org\u002F0000-0003-1751-9418",{"name":378,"orcid":9},"Yongkang Teng",{"name":380,"orcid":9},"Shenghao Ye",{"name":382,"orcid":383},"Menglei Wei","https:\u002F\u002Forcid.org\u002F0009-0004-3070-8023",{"name":385,"orcid":386},"Kun Yu","https:\u002F\u002Forcid.org\u002F0000-0002-0190-8702",{"name":388,"orcid":389},"Pingping Fang","https:\u002F\u002Forcid.org\u002F0000-0002-9911-2963",{"name":391,"orcid":9},"Jing Cao",{"name":393,"orcid":9},"Fanglin Zhu",{"name":395,"orcid":396},"Wenyu Zhang","https:\u002F\u002Forcid.org\u002F0000-0003-3322-9736",{"name":398,"orcid":399},"Ting Sun","https:\u002F\u002Forcid.org\u002F0000-0002-9387-4852",{"name":401,"orcid":9},"Min Jiang",{"tldr":403,"method":404,"finding":405,"direction":86,"opportunity":406},"提出CBAM-UNet分割与相机配置优化结合的水稻多视角三维重建流程，兼顾精度、完整性与效率。","Metatlas V1多视角平台、COLMAP+OpenMVS、CBAM-UNe","最优6相机配置将最近邻距离降至0.037–0.066 cm，保留73–76%点云，重建耗时减少约70","可将相机配置优化策略迁移到其他旋转多视角平台，并探索自适应配置与分割模型联合优化。","2026-09-25T23:30:25.303016Z",{"id":409,"title":410,"url":411,"summary":412,"summary_zh":413,"content":9,"source_name":324,"source_url":411,"published_at":414,"category":12,"cover_url":9,"hotness":204,"is_selected":14,"score":415,"score_detail":416,"sources":419,"tags":421,"primary_topic":423,"secondary_topics":429,"classification":431,"search_phrases":460,"slug":463,"view_count":223,"doi":464,"paper":465,"created_at":498},3165,"Explainable growth stage classification of cacao (Theobroma cacao L.) leaves and key feature visualization using vision transformer and transfer learning","https:\u002F\u002Fdoi.org\u002F10.3389\u002Ffpls.2026.1885906","Accurate and objective plant phenotyping is crucial for optimizing agricultural practices, understanding plant development, and enabling rapid responses to environmental changes. Traditional methods, often relying on visual observation, can be subjective, time-consuming, and may overlook subtle but important differences. This study demonstrates the power of combining digital imaging with deep learning to classify plant material with high accuracy, even when visual differences are minimal. We focused on differentiating between stage D (light green) and stage E (dark green) leaves of cacao ( Theobroma cacao L. ), which are visually very similar in size and overall structure. Using cleared and stained leaves of the SCA 6 genotype to highlight the venation network, we trained a Vision Transformer (ViT) model, a deep learning architecture, on image patches. At the patch level, the model achieved an overall accuracy of 97.03% on an independent test set, with a recall of 96.0% for stage D and 98.3% for stage E. At the whole-leaf level, majority voting correctly classified 14 of 15 independent test leaves (93.3%). Attention maps indicated that image regions containing the midrib and primary lateral veins contributed strongly to classification. These attention maps identify discriminative image regions, but they do not by themselves determine the biological mechanism underlying the signal. The major-vein signal may reflect developmental differences in vein-associated structure, stage-associated differences in Safranin O uptake or optical density, tissue thickness, or a combination of these factors. Because vascular anatomy, lignification, hydraulic conductance, phloem loading, and source–sink status were not directly measured, these mechanisms are treated as hypotheses requiring future anatomical, histochemical, and physiological validation. Thus, this study provides a proof-of-concept for interpretable image-based classification of stage D and stage E leaves within greenhouse-grown SCA 6 cacao. Extension to other cacao genotypes, field-grown plants, independent seasons, staining batches, stress detection, species identification, genotype discrimination, or precision-agriculture deployment will require external validation.","准确、客观的植物表型分析对于优化农业实践、理解植物发育以及快速响应环境变化至关重要。传统方法通常依赖视觉观察，可能具有主观性、耗时，并且可能忽略细微但重要的差异。本研究展示了将数字成像与深度学习相结合，即使在视觉差异极小的情况下，也能以高精度对植物材料进行分类。我们聚焦于区分可可（Theobroma cacao L.）的D期（浅绿色）和E期（深绿色）叶片，这些叶片在大小和整体结构上视觉上非常相似。利用SCA 6基因型的透明染色叶片以突出脉序网络，我们在图像块上训练了Vision Transformer（ViT）模型，一种深度学习架构。在图像块水平上，该模型在独立测试集上达到了97.03%的总体准确率，D期的召回率为96.0%，E期为98.3%。在整叶水平上，多数投票正确分类了15片独立测试叶片中的14片（93.3%）。注意力图表明，包含中脉和初级侧脉的图像区域对分类贡献显著。这些注意力图识别了具有判别力的图像区域，但它们本身并不能确定信号背后的生物学机制。主脉信号可能反映了脉相关结构的发育差异、番红O摄取或光密度的阶段相关差异、组织厚度，或这些因素的组合。由于未直接测量维管解剖结构、木质化、水力导度、韧皮部装载和源–库状态，这些机制被视为假设，需要未来的解剖学、组织化学和生理学验证。因此，本研究为温室种植的SCA 6可可中D期和E期叶片的可解释图像分类提供了概念验证。扩展到其他可可基因型、田间种植植株、独立季节、染色批次、胁迫检测、物种鉴定、基因型区分或精准农业部署将需要外部验证。","2026-09-21T00:00:00Z",71,{"impact":17,"substance":103,"depth":417,"authority":20,"freshness":21,"relevant":22,"comment":418},17,"方法新颖、数据可靠的可解释作物表型概念验证研究，但属实验室小样本，产业影响有限。",[420],{"name":324,"url":411},[29,30,274,422,31],"可可",{"code":424,"slug":425,"name":426,"path":427,"parent_code":428},"B3","tea-coffee","茶叶与咖啡","\u002Ftopic\u002Fhorticulture\u002Ftea-coffee","B",[430],{"code":47,"slug":215,"name":216,"path":217,"parent_code":118},{"status":121,"locked":14,"version":37,"method":38,"evidence":432},{"candidates":433,"agricultural_evidence":56},[434,439,444,451,456],{"code":424,"score":125,"supported":56,"title":435,"abstract":437,"tags":438},[436],"cacao",[422,436],[422],{"code":47,"score":104,"supported":56,"title":440,"abstract":441,"tags":443},[],[274,442],"deep learning",[30,274],{"code":445,"score":91,"supported":56,"title":446,"abstract":447,"tags":450},"B7",[],[448,449],"温室","greenhouse",[],{"code":42,"score":150,"supported":56,"title":452,"abstract":453,"tags":455},[],[454],"精准农业",[29],{"code":52,"score":22,"supported":14,"title":457,"abstract":458,"tags":459},[],[],[31],[461,462],"可可 叶片 生长阶段 分类","Vision Transformer 作物表型","可可叶片生长阶段分类-3165","10.3389\u002Ffpls.2026.1885906",{"doi":464,"openalex_id":466,"authors":467,"venue":324,"cited_by_count":81,"oa_url":411,"card":493,"direction":86,"ingested_from":88},"W7213901640",[468,470,473,475,477,479,482,485,488,490],{"name":469,"orcid":9},"Ezekiel Ahn",{"name":471,"orcid":472},"Eun-Sung Park","https:\u002F\u002Forcid.org\u002F0000-0001-6826-2865",{"name":474,"orcid":9},"Moon S. Kim",{"name":476,"orcid":9},"Hangi Kim",{"name":478,"orcid":9},"Lalit M. Kandpal",{"name":480,"orcid":481},"Sunchung Park","https:\u002F\u002Forcid.org\u002F0000-0002-7398-9476",{"name":483,"orcid":484},"Seunghyun Lim","https:\u002F\u002Forcid.org\u002F0000-0003-3023-4863",{"name":486,"orcid":487},"Lyndel W. Meinhardt","https:\u002F\u002Forcid.org\u002F0000-0001-8299-2629",{"name":489,"orcid":9},"Byoung-Kwan Cho",{"name":491,"orcid":492},"Insuck Baek","https:\u002F\u002Forcid.org\u002F0000-0003-1044-349X",{"tldr":494,"method":495,"finding":496,"direction":86,"opportunity":497},"用ViT和迁移学习对可可叶D、E期进行可解释分类，准确率达97%。","透明染色叶片图像块训练ViT，注意力图可视化关键区域。","模型准确区分D\u002FE期，中脉和主侧脉区域贡献最大。","可扩展到多基因型、田间、胁迫检测，并验证脉信号生物学机制。","2026-09-22T23:30:19.972949Z",{"id":500,"title":501,"url":502,"summary":503,"summary_zh":504,"content":9,"source_name":99,"source_url":502,"published_at":414,"category":12,"cover_url":9,"hotness":204,"is_selected":14,"score":505,"score_detail":506,"sources":508,"tags":510,"primary_topic":512,"secondary_topics":513,"classification":516,"search_phrases":538,"slug":541,"view_count":282,"doi":542,"paper":543,"created_at":559},3164,"Spatiotemporal Deep Learning for Rice Plant Height Estimation from Multi-Temporal UAV RGB Imagery","https:\u002F\u002Fdoi.org\u002F10.3390\u002Fagriculture16182034","Accurate plant height estimation is important for monitoring crop growth and supporting precision agricultural management. Manual measurements are labor-intensive, while LiDAR-based methods are expensive and require complex processing. UAV photogrammetry provides a lower-cost alternative but remains challenging in flooded rice paddies because of canopy deformation and difficulties in terrain extraction. This study proposes Rice-STNet, a spatiotemporal deep learning framework for end-to-end rice plant height estimation using multi-temporal UAV RGB imagery. Rice-STNet integrates a convolutional neural network for spatial feature extraction, Time2Vec for temporal encoding, and a gated recurrent unit network for modeling temporal dependencies across observation dates. The framework was evaluated using field data collected from rice paddies over two growing seasons. Rice-STNet achieved an R2 of 0.97, a root mean squared error of 1.97 cm, and a mean absolute error of 1.14 cm. It outperformed random forest, support vector regression, a CNN-only baseline, and a UAV photogrammetry-based point-cloud approach. In addition, the framework generated high-resolution plant height maps for field-scale analysis of spatial growth variability. These results underscore the importance of jointly modeling spatial and temporal characteristics for continuously evolving crop traits. The proposed framework offers an accurate, scalable, and non-destructive solution for large-scale crop phenotyping and precision agriculture.","准确的株高估算对于监测作物生长和支持精准农业管理具有重要意义。人工测量劳动强度大，而基于激光雷达（LiDAR）的方法成本高昂且需要复杂的处理。无人机摄影测量提供了一种成本较低的替代方案，但在淹水稻田中仍面临挑战，原因在于冠层变形和地形提取困难。本研究提出了Rice-STNet，一种时空深度学习框架，用于利用多时相无人机RGB影像进行端到端水稻株高估算。Rice-STNet集成了用于空间特征提取的卷积神经网络、用于时间编码的Time2Vec，以及用于建模观测日期之间时间依赖关系的门控循环单元网络。该框架利用两个生长季从稻田采集的田间数据进行了评估。Rice-STNet取得了R²为0.97、均方根误差为1.97 cm、平均绝对误差为1.14 cm的结果。其性能优于随机森林、支持向量回归、仅使用CNN的基线方法以及基于无人机摄影测量的点云方法。此外，该框架生成了高分辨率株高图，用于田块尺度空间生长变异性分析。这些结果凸显了联合建模空间与时间特征对于持续变化的作物性状的重要性。所提出的框架为大规模作物表型分析和精准农业提供了一种准确、可扩展且非破坏性的解决方案。",80,{"impact":19,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":507},"提出时空深度学习框架Rice-STNet，用多时相无人机RGB影像实现水稻株高高精度估算，方法新颖、数据跨两个生长季，对作物表型与精准农业有实用价值。",[509],{"name":99,"url":502},[29,30,332,511,31],"遥感",{"code":335,"slug":336,"name":332,"path":337,"parent_code":338},[514,515],{"code":47,"slug":215,"name":216,"path":217,"parent_code":118},{"code":52,"slug":115,"name":116,"path":117,"parent_code":118},{"status":121,"locked":14,"version":37,"method":38,"evidence":517},{"candidates":518,"agricultural_evidence":56},[519,524,528,532],{"code":335,"score":19,"supported":56,"title":520,"abstract":521,"tags":523},[336],[332,522,336],"稻田",[332],{"code":47,"score":125,"supported":56,"title":525,"abstract":526,"tags":527},[442],[274,442],[30],{"code":52,"score":104,"supported":56,"title":529,"abstract":530,"tags":531},[],[31,132],[31,511],{"code":42,"score":533,"supported":56,"title":534,"abstract":535,"tags":537},7,[],[454,536],"precision agriculture",[29],[539,540],"无人机 RGB 水稻株高","Rice-STNet 水稻表型","无人机RGB水稻株高-3164","10.3390\u002Fagriculture16182034",{"doi":542,"openalex_id":544,"authors":545,"venue":99,"cited_by_count":81,"oa_url":502,"card":554,"direction":86,"ingested_from":88},"W7213887432",[546,549,551],{"name":547,"orcid":548},"Weiguo Wang","https:\u002F\u002Forcid.org\u002F0009-0003-4028-9363",{"name":550,"orcid":9},"Noboru Noguchi",{"name":552,"orcid":553},"Liangliang Yang","https:\u002F\u002Forcid.org\u002F0000-0002-5055-3987",{"tldr":555,"method":556,"finding":557,"direction":86,"opportunity":558},"提出Rice-STNet时空深度学习框架，用多时相无人机RGB影像估算水稻株高。","CNN提取空间特征，Time2Vec编码时间，GRU建模时序依赖，两季稻田数据验","R²达0.97、RMSE 1.97cm，优于随机森林、SVR、纯CNN及点云方法。","可迁移至其他作物与多源遥感融合，探索轻量化模型及实时田间部署。","2026-09-22T23:30:18.545958Z"]