[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-797":3,"related-797":45},{"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":23,"tags":25,"primary_topic":8,"secondary_topics":31,"classification":32,"search_phrases":39,"slug":42,"view_count":43,"doi":8,"paper":8,"created_at":44},797,"[预印本]HarvestPoint-ACT:遮挡条件下机器人水果采摘的显式目标选择与采摘点条件化(原标题:HarvestPoint-ACT: Explicit Target Selection and Harvest-Point Conditioning for Robotic Fruit Harvesting under Occlusion)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.18446","端到端模仿学习避开了手工制作接近抓取的机器人运动,但策略仍需决定采摘哪个水果以及在何处闭合夹爪。遮挡会使策略在采摘过程中丢失所选水果,正确的闭合点也难以仅从像素推断。本研究提出HarvestPoint-ACT,在感知中使两个决策显式化并提供给策略:实例分割前端带关键点分支为每个可见水果预测掩码和采摘点;调度器按遮挡和移动距离对检测候选排序并选择一个目标。每次尝试后重新检测和重新排序候选。所选水果被编码为8维状态,送入动作分块Transformer。在冠层模拟装置上成功率88%,重遮挡下75%。",null,"[View PDF](https:\u002F\u002Farxiv.org\u002Fpdf\u002F2608.18446)[HTML (experimental)](https:\u002F\u002Farxiv.org\u002Fhtml\u002F2608.18446v1)\n\n> Abstract:End-to-end imitation learning avoids hand-made robot motion for approaching and grasping, but the policy must still decide which fruit to pick and where to close the gripper. Occlusion can make the policy lose the selected fruit during harvesting, and the correct closing point is difficult to infer from pixels alone. This paper presents HarvestPoint-ACT, which makes both decisions explicit in perception and provides them to the policy. An instance segmentation front end with a keypoint branch predicts a mask and a harvest point for each visible fruit, where the harvest point specifies the location to close the gripper. A scheduler ranks detected candidates by occlusion and travel distance and selects one target. After each attempt, it redetects and reranks the candidates because the canopy may have changed. The selected fruit is encoded for an action chunking transformer as an eight-dimensional state, containing the absolute harvest point, the vector from the gripper to that point, a validity flag, and a confidence score. When the selected fruit is temporarily undetected, the system retains the last harvest point estimate in the robot base frame and marks it as stale, and aborts the attempt if the loss persists. On a canopy mock-up, HarvestPoint-ACT achieves a success rate of 88%, and of 75% under heavy occlusion.\n\nComments:Submit to IEEE ROBIO 2026\nSubjects:Robotics (cs.RO)\nCite as:[arXiv:2608.18446](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.18446) [cs.RO]\n(or [arXiv:2608.18446v1](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.18446v1) [cs.RO] for this version)\n[https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.18446](https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.18446)\n\narXiv-issued DOI via DataCite (pending registration)\n\n## Submission history\n\nFrom: Hao Chen [[view email](https:\u002F\u002Farxiv.org\u002Fshow-email\u002F2eea8537\u002F2608.18446)] \n\n**[v1]** Wed, 19 Aug 2026 02:21:27 UTC (7,450 KB)","arXiv:2608.18446 2026-08-19 [预印本]","2026-08-18T18:21:00Z","论文",10,false,78,{"impact":17,"substance":18,"depth":17,"authority":19,"freshness":20,"relevant":21,"comment":22},18,22,12,8,1,"针对遮挡条件下机器人采摘的显式目标选择与采摘点条件化方法，具有较高技术价值，但属于细分领域进展，影响范围有限。",[24],{"name":10,"url":6},[26,27,28,29,30],"预印本","机器人采摘","遮挡条件","HarvestPoint-ACT","动作分块Transformer",[],{"status":33,"locked":14,"version":34,"method":35,"evidence":36},"needs_review","2026-10-22.2","rules",{"candidates":37,"agricultural_evidence":14,"reason":38},[],"农业相关性需要复核",[40,41],"机器人采摘 动作分块Transformer 遮挡条件 预印本","机器人采摘 动作分块Transformer","机器人采摘动作分块Transformer遮挡条件预印本-797",2,"2026-08-21T00:05:57.773126Z",{"total":46,"page":21,"page_size":46,"items":47},6,[48,104,152,208,249,285],{"id":49,"title":50,"url":51,"summary":52,"summary_zh":8,"content":8,"source_name":53,"source_url":8,"published_at":54,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":55,"score_detail":56,"sources":59,"tags":61,"primary_topic":66,"secondary_topics":72,"classification":73,"search_phrases":88,"slug":91,"view_count":92,"doi":8,"paper":93,"created_at":103},2396,"[预印本]Multi-Task Learning for Sparsely-Labeled Time Series: A Case Study on Cold-Hardiness Modeling","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.09062","arXiv 2609.09062（2026-09-10）。研究针对葡萄抗寒性的时间序列过程建模问题（数据有限且标签时序稀疏），提出基于RNN的多任务学习方法。实验证明某些MTL架构优于单任务学习和现有科学模型，并在相关萌芽期预测任务上展现良好潜力。","arXiv | 2026-09-10","2026-09-10T00:00:00Z",66,{"impact":19,"substance":17,"depth":57,"authority":19,"freshness":20,"relevant":21,"comment":58},16,"预印本提出RNN多任务学习用于葡萄抗寒性稀疏标签时序建模，方法新颖但尚属早期研究，产业影响有限，可作为农业AI前沿动态收录。",[60],{"name":53,"url":51},[62,26,63,64,65],"农业人工智能","葡萄","抗寒育种","时间序列",{"code":67,"slug":68,"name":69,"path":70,"parent_code":71},"B2","fruit","果树","\u002Ftopic\u002Fhorticulture\u002Ffruit","B",[],{"status":74,"locked":14,"version":34,"method":35,"evidence":75},"classified",{"candidates":76,"agricultural_evidence":79},[77,83],{"code":67,"score":78,"supported":79,"title":80,"abstract":81,"tags":82},4,true,[],[63],[63],{"code":84,"score":21,"supported":14,"title":85,"abstract":86,"tags":87},"I2",[],[],[62],[89,90],"农业人工智能 抗寒育种 时间序列 预印本","农业人工智能 抗寒育种","农业人工智能抗寒育种时间序列预印本-2396",3,{"doi":8,"openalex_id":8,"authors":94,"venue":8,"cited_by_count":95,"oa_url":8,"card":96,"direction":100,"ingested_from":102},[],0,{"tldr":97,"method":98,"finding":99,"direction":100,"opportunity":101},"提出RNN多任务学习方法，用于标签稀疏的葡萄抗寒时间序列建模。","基于RNN的多任务学习，处理有限且时序稀疏的标签数据。","部分MTL架构优于单任务学习和现有科学模型，并利于萌芽期预测。","农业人工智能与决策模型","可探索多任务学习在更多稀疏标签农业时序场景（如病虫害、物候）中的迁移与泛化。","agent","2026-09-14T00:06:33.014967Z",{"id":105,"title":106,"url":107,"summary":108,"summary_zh":8,"content":109,"source_name":110,"source_url":8,"published_at":111,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":112,"score_detail":113,"sources":117,"tags":119,"primary_topic":128,"secondary_topics":134,"classification":135,"search_phrases":147,"slug":150,"view_count":92,"doi":8,"paper":8,"created_at":151},874,"Reinforcement Learning加Visual SLAM融合高光谱成像实现精准农业UAV作物水分胁迫自动检测:92%检测率vs固定路径61%,飞行时间减32%","https:\u002F\u002Fengrxiv.org\u002Fpreprint\u002Fview\u002F8004","Arsalan Iqbal在《engrXiv》预印本平台发布研究,提出一种基于强化学习引导的自主四旋翼无人机平台,融合机载视觉同步定位与建图(Visual SLAM)与推扫式高光谱成像载荷,在近实时构建地理参考的、冠层配准的作物水分胁迫指数(CWSI)地图。平台采用近端策略优化(PPO)训练的自适应采样策略,根据飞行中积累的证据将飞行时间和传感器驻留时间重新分配到新兴水分胁迫区域。在0.8公顷试验田模拟田间试验中,强化学习引导策略实现了92%的水分胁迫区域检测率(固定网格基线为61%),同时任务飞行时间减少约32%。研究还报告了SLAM衍生的冠层结构对CWSI精度的贡献消融研究、田间复杂性敏感性和全误差预算分析。注:本研究为预印本,未经实地验证。","Preprint\u002FVersion 1\n\n## ##article.authors##\n\n## DOI:\n\n[https:\u002F\u002Fdoi.org\u002F10.31224\u002F8004](https:\u002F\u002Fdoi.org\u002F10.31224\u002F8004)\n## Keywords:\n\n VSLAM, Hyperspectral, Remote Sensing, Crop Water-Stress Index, Reinforcement Learning, Proximal Policy Optimization, Adaptive Path Planning \n## Abstract\n\nLocalized soil-moisture deficits, that is, irregular sub-field patches where crops experience water stress well before visible wilting, are a leading cause of yield variability in row-crop agriculture. These zones are difficult to detect at the spatial resolution and revisit frequency required for timely irrigation response. This paper presents a reinforcement-learningguided autonomous quadrotor unmanned aerial vehicle (UAV) platform that fuses onboard Visual Simultaneous Localization and Mapping (Visual SLAM) with a pushbroom hyperspectral imaging payload to construct georeferenced, canopy-registered maps of a Crop Water-Stress Index (CWSI) in near real time. Rather than flying a fixed lawnmower survey, the platform is guided by an adaptive-sampling policy trained with Proximal Policy Optimization (PPO) that reallocates flight time and sensor dwell toward regions of emerging water stress as evidence accumulates mid-flight. We present the complete engineering pipeline: airframe and sensor design, a keyframe-based Visual SLAM front and back end that provides centimeter-scale geolocation without continuous reliance on Real-Time Kinematic (RTK) GNSS lock, a hyperspectral preprocessing and spectralindex chain (NDVI, NDRE, NDWI\u002FNDMI) used to derive CWSI through a learned regression, the partially observable Markov Decision Process (POMDP) formulation and reward shaping used to train the sampling policy, and the fused system architecture tying these subsystems together. In simulated field trials over a 0.8-hectare test plot, the reinforcement-learning-guided policy achieved a 92% water-stress-zone detection rate versus 61% for a fixed-grid baseline, while reducing mission flight time by approximately 32%. We further report an ablation study isolating the contribution of SLAM-derived canopy structure to CWSI accuracy, a sensitivity analysis across field complexity, and a full error budget for the fused pipeline. We close with a discussion of validation limitations, broader scientific and agricultural impact, and a roadmap toward multi-UAV fleet deployment for whole-farm monitoring\n\n## Posted\n\n2026-08-19","engrXiv preprint 2026年8月19日","2026-08-19T00:00:00Z",77,{"impact":17,"substance":18,"depth":114,"authority":13,"freshness":115,"relevant":21,"comment":116},20,7,"论文提出RL+SLAM+高光谱融合的精准农业无人机方案，检测率提升显著，方法新颖，数据详实，但为预印本且影响范围限于细分领域。",[118],{"name":110,"url":107},[120,26,121,122,123,124,125,126,127],"无人机","精准农业","强化学习","高光谱成像","Visual SLAM","作物水分胁迫","PPO","engrXiv",{"code":129,"slug":130,"name":131,"path":132,"parent_code":133},"I1","smart-agriculture","智慧农业总论","\u002Ftopic\u002Fdigital-agriculture\u002Fsmart-agriculture","I",[],{"status":74,"locked":14,"version":34,"method":35,"evidence":136},{"candidates":137,"agricultural_evidence":79},[138,143],{"code":129,"score":139,"supported":79,"title":140,"abstract":141,"tags":142},9,[121],[],[121],{"code":84,"score":21,"supported":14,"title":144,"abstract":145,"tags":146},[],[],[122],[148,149],"作物水分胁迫 高光谱成像 强化学习 精准农业","作物水分胁迫 高光谱成像","作物水分胁迫高光谱成像强化学习精准农业-874","2026-08-23T00:08:13.490346Z",{"id":153,"title":154,"url":155,"summary":156,"summary_zh":8,"content":157,"source_name":158,"source_url":8,"published_at":159,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":160,"score_detail":161,"sources":163,"tags":165,"primary_topic":170,"secondary_topics":176,"classification":183,"search_phrases":203,"slug":206,"view_count":78,"doi":8,"paper":8,"created_at":207},796,"[预印本]自主农业拖拉机:面向精准稻作的集成杂草检测与LiDAR导航系统(原标题:Autonomous Agricultural Tractor: Integrated Weed Detection and LiDAR Navigation for Precision Paddy Farming)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.19004","稻田定点杂草管理相比传统广域喷药可大幅减少除草剂用量,但现场部署受三大挑战制约:冠层下GNSS退化的稳健作物行导航、稻与形态多样化杂草的实时视觉区分、将稻误识为杂草的不可逆错误代价。本研究提出AgriNav,围绕四个ROS耦合模块构建集成自主拖拉机系统:用于稻检测的WeedDet PyTorch重实现、并行轻量级168万参数CNN-FPN变体(带非对称类加权)、通过硬编码置信门否决保护稻类的逆逻辑判别模块,以及融合GNSS\u002FIMU\u002F轮式里程计的6状态恒定速度转向率扩展卡尔曼滤波。主要系统级贡献是LiDAR-相机融合桥,利用导航LiDAR做ROI约束、世界坐标投影、地面平面滤波和双向置信融合,无额外硬件成本。","[View PDF](https:\u002F\u002Farxiv.org\u002Fpdf\u002F2608.19004)[HTML (experimental)](https:\u002F\u002Farxiv.org\u002Fhtml\u002F2608.19004v1)\n\n> Abstract:Site-specific weed management in paddy farming offers substantial reductions in herbicide use over conventional broadcast spraying, but field deployment has been limited by three persistent challenges: robust crop-row navigation under canopy where GNSS degrades, real-time visual discrimination between rice and morphologically diverse weeds, and the asymmetric cost of misclassifying rice as weed, which is irreversible. \n> \n> This paper presents AgriNav, an integrated autonomous tractor system built around four ROS-coupled modules: a custom PyTorch reimplementation of WeedDet for rice detection, a parallel lightweight 1.68M-parameter CNN-FPN variant with asymmetric class weighting, an inverted-logic discrimination module that protects the rice class through a hardcoded confidence-gate veto, and a 6-state constant-velocity-turn-rate Extended Kalman Filter fusing GNSS, IMU, and wheel odometry with three-level outage bridging. Our primary system-level contribution is a four-mechanism LiDAR-camera fusion bridge that uses the navigation LiDAR for region-of-interest constraint, world-coordinate projection, ground-plane filtering, and bidirectional confidence fusion at zero additional hardware cost. \n> \n> Simulation experiments demonstrate continuous position tracking through a 20-second GNSS outage, crop row detection confidence above 0.9 throughout operation, and rice-detection confidences from 0.32 to 0.95 across paddy, aerial, and post-flood imagery. The LiDAR ROI constraint reduces detection inference region by an estimated 30 to 50 percent.\n\nSubjects:Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)\nCite as:[arXiv:2608.19004](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.19004) [cs.RO]\n(or [arXiv:2608.19004v1](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.19004v1) [cs.RO] for this version)\n[https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.19004](https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.19004)\n\narXiv-issued DOI via DataCite (pending registration)\n\n## Submission history\n\nFrom: Sudip Dhakal [[view email](https:\u002F\u002Farxiv.org\u002Fshow-email\u002Ff6fe364c\u002F2608.19004)] \n\n**[v1]** Wed, 19 Aug 2026 15:04:22 UTC (1,887 KB)","arXiv:2608.19004 2026-08-19 [预印本]","2026-08-19T07:04:00Z",64,{"impact":19,"substance":17,"depth":57,"authority":13,"freshness":20,"relevant":21,"comment":162},"面向精准稻作的自主拖拉机系统，集成杂草检测与LiDAR导航，技术细节丰富，但属预印本且应用场景特定，影响范围有限。",[164],{"name":158,"url":155},[26,166,167,168,169],"自主拖拉机","LiDAR导航","杂草检测","稻田",{"code":171,"slug":172,"name":173,"path":174,"parent_code":175},"A1","rice","水稻","\u002Ftopic\u002Ffield-crops\u002Frice","A",[177],{"code":178,"slug":179,"name":180,"path":181,"parent_code":182},"F5","machinery","农机装备与机械化","\u002Ftopic\u002Finputs-equipment\u002Fmachinery","F",{"status":74,"locked":14,"version":34,"method":35,"evidence":184},{"candidates":185,"agricultural_evidence":79},[186,192,197],{"code":171,"score":187,"supported":79,"title":188,"abstract":190,"tags":191},11,[189],"paddy",[169],[],{"code":178,"score":20,"supported":79,"title":193,"abstract":195,"tags":196},[194],"tractor",[],[],{"code":198,"score":92,"supported":79,"title":199,"abstract":200,"tags":202},"F4",[],[201],"除草",[],[204,205],"自主拖拉机 杂草检测 预印本 LiDAR导航","自主拖拉机 杂草检测","自主拖拉机杂草检测预印本LiDAR导航-796","2026-08-21T00:05:57.744668Z",{"id":209,"title":210,"url":211,"summary":212,"summary_zh":8,"content":213,"source_name":214,"source_url":8,"published_at":215,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":95,"score_detail":216,"sources":218,"tags":220,"primary_topic":224,"secondary_topics":228,"classification":229,"search_phrases":244,"slug":247,"view_count":92,"doi":8,"paper":8,"created_at":248},710,"UltraIR:从分子到复杂样品的红外光谱化学感知与分析仿真到现实迁移学习 [预印本]","https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.13341","中国作者团队发布1 亿参数红外光谱基础模型 UltraIR,基于约 6,000 万模拟光谱预训练并融合谱图重建、分子指纹对齐与官能团预测;在官能团预测、混合物定量、细菌分类、药材产地鉴别、微塑料分类及土壤属性预测等任务上均超越传统机器学习与任务专用深度学习,少样本与跨实验室零样本迁移能力显著,为农业土壤属性检测、药材溯源与农产品品质化学感知提供数据高效新路径。[预印本,未经同行评审]","Authors:[Yusen Tan](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Tan,+Y), [Yixuan Chen](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Chen,+Y), [Zheng Fang](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Fang,+Z), [Pan Liu](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Liu,+P), [Yifan Li](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Li,+Y), [Qinyu Guo](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Guo,+Q), [Zhedong Lin](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Lin,+Z), [Yuqiang Li](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Li,+Y), [Xiangxiang Zeng](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Zeng,+X), [Tong Wang](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Wang,+T), [Jun Xia](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Xia,+J)\n\n[View PDF](https:\u002F\u002Farxiv.org\u002Fpdf\u002F2608.13341)[HTML (experimental)](https:\u002F\u002Farxiv.org\u002Fhtml\u002F2608.13341v2)\n\n> Abstract:Infrared (IR) spectroscopy is widely used for chemical sensing, but extracting reliable chemical information from spectra remains challenging. Conventional interpretation is labor-intensive, relies on prior knowledge and reference spectra, and is difficult to scale, whereas most machine-learning methods are tailored to individual tasks or datasets, require large labeled training sets, and transfer poorly across analytical objectives and experimental datasets. Here we introduce UltraIR, a foundation model for IR spectroscopy with more than 100 million parameters that enables simulation-to-real transfer learning for chemical sensing and analysis from molecules to complex samples. UltraIR is pretrained on approximately 60 million simulated IR spectra using spectral reconstruction, molecular fingerprint similarity alignment, and functional-group prediction, then adapted to downstream objectives with task-specific labels or targets. Across functional-group prediction, molecular structure elucidation, physicochemical property prediction, mixture-component identification and quantification, bacterial classification, medicinal-herb geographic origin traceability and constituent quantification, microplastics classification, and soil property prediction, UltraIR outperforms conventional machine-learning and task-specific deep-learning baselines. It performs strongly with limited labeled experimental spectra and in zero-shot inference for the same analytical task across Fourier-transform infrared spectrometers and laboratories, providing a route to adaptable, data-efficient chemical sensing from complex real-world samples.\n\nSubjects:Machine Learning (cs.LG); Artificial Intelligence (cs.AI)\nCite as:[arXiv:2608.13341](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.13341) [cs.LG]\n(or [arXiv:2608.13341v2](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.13341v2) [cs.LG] for this version)\n[https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.13341](https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.13341)\n\narXiv-issued DOI via DataCite\n\n## Submission history\n\nFrom: Yusen Tan [[view email](https:\u002F\u002Farxiv.org\u002Fshow-email\u002F2fb6ad01\u002F2608.13341)] \n\n**[[v1]](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.13341v1)** Thu, 13 Aug 2026 15:11:50 UTC (7,895 KB)\n\n**[v2]** Fri, 14 Aug 2026 03:01:37 UTC (7,895 KB)","arXiv:2608.13341 [cs.LG] 2026-08-14","2026-08-14T00:00:00Z",{"impact":95,"substance":95,"depth":95,"authority":95,"freshness":95,"relevant":95,"comment":217},"该论文主题为红外光谱化学感知与分析，属于化学与机器学习交叉领域，与农业信息化、智慧农业等主题无直接关联，不满足相关性门槛。",[219],{"name":214,"url":211},[62,26,221,222,223],"大模型","红外光谱","土壤检测",{"code":84,"slug":225,"name":226,"path":227,"parent_code":133},"agricultural-ai","农业AI与大模型","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-ai",[],{"status":74,"locked":14,"version":34,"method":35,"evidence":230},{"candidates":231,"agricultural_evidence":79},[232,238],{"code":84,"score":20,"supported":79,"title":233,"abstract":234,"tags":237},[],[235,236],"深度学习","机器学习",[62,221],{"code":239,"score":78,"supported":79,"title":240,"abstract":241,"tags":243},"H2",[],[242],"土壤",[223],[245,246],"农业人工智能 土壤检测 红外光谱 大模型","农业人工智能 土壤检测","农业人工智能土壤检测红外光谱大模型-710","2026-08-18T23:59:19.229607Z",{"id":250,"title":251,"url":252,"summary":253,"summary_zh":8,"content":254,"source_name":255,"source_url":8,"published_at":256,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":257,"score_detail":258,"sources":260,"tags":262,"primary_topic":267,"secondary_topics":268,"classification":269,"search_phrases":280,"slug":283,"view_count":46,"doi":8,"paper":8,"created_at":284},415,"AgriField-40K: Adapting Vision Models to Agriculture With Efficient Continual Pretraining","https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984","Vasileios Tzouras、Paraskevas Pegios、Lazaros Nalpantidis于arXiv(2026-08-08)发表预印本。研究引入AgriField-40K,一个源自17个公开数据集的以田野为中心的农业数据集,涵盖多样化作物、杂草和田间条件;并提出AgriMAE,一种参数高效的持续预训练基线,通过仅训练轻量适配器来适配在自然图像上预训练的掩码自动编码器。AgriMAE在下游任务上一致提升性能,使用多达9倍更少的可训练参数即可匹配甚至超越全量微调,证明AgriField-40K是农业视觉持续预训练实用的资源。为精准农业模型开发提供低成本高效率的迁移学习路径,对农业视觉应用具有显著价值。","[Skip to main content](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984#content)[](https:\u002F\u002Farxiv.org\u002FIgnoreMe)[![Image 1: archive](https:\u002F\u002Farxiv.org\u002Fstatic\u002Fbase\u002F1.0.1\u002Fimages\u002Farxiv-logo-primary-light.svg)](https:\u002F\u002Farxiv.org\u002F)[Search](https:\u002F\u002Farxiv.org\u002Fsearch)[Submit](https:\u002F\u002Farxiv.org\u002Fuser\u002Fcreate)[Donate](https:\u002F\u002Finfo.arxiv.org\u002Fabout\u002Fdonate.html)[Log in](https:\u002F\u002Farxiv.org\u002Flogin)\n\nSearch arXiv \n\n Press Enter to search · [Advanced search](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fadvanced)\n\n# Computer Science > Computer Vision and Pattern Recognition\n\n**arXiv:2608.07984** (cs) \n\n [Submitted on 8 Aug 2026]\n\n# Title:AgriField-40K: Adapting Vision Models to Agriculture With Efficient Continual Pretraining\n\nAuthors:[Vasileios Tzouras](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Tzouras,+V), [Paraskevas Pegios](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Pegios,+P), [Lazaros Nalpantidis](https:\u002F\u002Farxiv.org\u002Fsearch\u002Fcs?searchtype=author&query=Nalpantidis,+L)\n\nView a PDF of the paper titled AgriField-40K: Adapting Vision Models to Agriculture With Efficient Continual Pretraining, by Vasileios Tzouras and 2 other authors\n\n[View PDF](https:\u002F\u002Farxiv.org\u002Fpdf\u002F2608.07984)[HTML (experimental)](https:\u002F\u002Farxiv.org\u002Fhtml\u002F2608.07984v1)\n> Abstract:Field-based agricultural computer vision is important for precision agriculture, yet it largely depends on expensive annotations and costly adaptation of large pretrained models. We introduce AgriField-40K, a field-centric dataset curated from 17 public resources and covering diverse crops, weeds, and field conditions. Building on this, we present AgriMAE, a parameter-efficient continual pretraining baseline that adapts a masked autoencoder pretrained on natural images by training only lightweight adapters. We further explore semantic feature reconstruction as an alternative pretraining objective and evaluate transfer across multiple tasks. AgriMAE consistently improves downstream performance and can match or even outperform full fine-tuning while using up to $9\\times$ fewer trainable parameters, showing that AgriField-40K is a practical resource for continual pretraining in agricultural vision. Project page: [this https URL](https:\u002F\u002Fdtu-pas.github.io\u002Fagrifield40k\u002F)\n\nComments:Accepted at the Computer Vision in Plant Phenotyping and Agriculture (CVPPA) Workshop at the European Conference on Computer Vision (ECCV) 2026\nSubjects:Computer Vision and Pattern Recognition (cs.CV)\nCite as:[arXiv:2608.07984](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984) [cs.CV]\n(or [arXiv:2608.07984v1](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984v1) [cs.CV] for this version)\n[https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.07984](https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.07984)\n\nFocus to learn more\n\n arXiv-issued DOI via DataCite (pending registration)\n\n## Submission history\n\n From: Vasileios Tzouras [[view email](https:\u002F\u002Farxiv.org\u002Fshow-email\u002F09ea92b8\u002F2608.07984)] \n\n**[v1]** Sat, 8 Aug 2026 07:38:01 UTC (24,651 KB)\n\n[](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984)Full-text links:\n## Access Paper:\n\n View a PDF of the paper titled AgriField-40K: Adapting Vision Models to Agriculture With Efficient Continual Pretraining, by Vasileios Tzouras and 2 other authors\n\n*   [View PDF](https:\u002F\u002Farxiv.org\u002Fpdf\u002F2608.07984)\n*   [HTML (experimental)](https:\u002F\u002Farxiv.org\u002Fhtml\u002F2608.07984v1)\n*   [TeX Source](https:\u002F\u002Farxiv.org\u002Fsrc\u002F2608.07984)\n\n[![Image 2: license icon](https:\u002F\u002Farxiv.org\u002Ficons\u002Flicenses\u002Fby-4.0.png)view license](http:\u002F\u002Fcreativecommons.org\u002Flicenses\u002Fby\u002F4.0\u002F \"Rights to this article\")\n\n### Current browse context:\n\ncs.CV\n\n[\u003Cprev](https:\u002F\u002Farxiv.org\u002Fprevnext?id=2608.07984&function=prev&context=cs.CV \"previous in cs.CV (accesskey p)\") | [next>](https:\u002F\u002Farxiv.org\u002Fprevnext?id=2608.07984&function=next&context=cs.CV \"next in cs.CV (accesskey n)\")\n\n[new](https:\u002F\u002Farxiv.org\u002Flist\u002Fcs.CV\u002Fnew) | [recent](https:\u002F\u002Farxiv.org\u002Flist\u002Fcs.CV\u002Frecent) | [2026-08](https:\u002F\u002Farxiv.org\u002Flist\u002Fcs.CV\u002F2026-08)\n\n Change to browse by: \n\n[cs](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984?context=cs)\n\n### References & Citations\n\n*   [NASA ADS](https:\u002F\u002Fui.adsabs.harvard.edu\u002Fabs\u002FarXiv:2608.07984)\n*   [Google Scholar](https:\u002F\u002Fscholar.google.com\u002Fscholar_lookup?arxiv_id=2608.07984)\n*   [Semantic Scholar](https:\u002F\u002Fapi.semanticscholar.org\u002FarXiv:2608.07984)\n\nexport BibTeX citation Loading...\n\n## BibTeX formatted citation\n\n×\n\nData provided by: [](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984)\n\n### Bookmark\n\n[![Image 3: BibSonomy](https:\u002F\u002Farxiv.org\u002Fstatic\u002Fbrowse\u002F0.3.4\u002Fimages\u002Ficons\u002Fsocial\u002Fbibsonomy.png)](http:\u002F\u002Fwww.bibsonomy.org\u002FBibtexHandler?requTask=upload&url=https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984&description=AgriField-40K:%20Adapting%20Vision%20Models%20to%20Agriculture%20With%20Efficient%20Continual%20Pretraining \"Bookmark on BibSonomy\")[![Image 4: Reddit](https:\u002F\u002Farxiv.org\u002Fstatic\u002Fbrowse\u002F0.3.4\u002Fimages\u002Ficons\u002Fsocial\u002Freddit.png)](https:\u002F\u002Freddit.com\u002Fsubmit?url=https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984&title=AgriField-40K:%20Adapting%20Vision%20Models%20to%20Agriculture%20With%20Efficient%20Continual%20Pretraining \"Bookmark on Reddit\")\n\nBibliographic Tools \n\n# Bibliographic and Citation Tools\n\n- [x] Bibliographic Explorer Toggle \n\nBibliographic Explorer _([What is the Explorer?](https:\u002F\u002Finfo.arxiv.org\u002Flabs\u002Fshowcase.html#arxiv-bibliographic-explorer))_\n\n- [x] Connected Papers Toggle \n\nConnected Papers _([What is Connected Papers?](https:\u002F\u002Fwww.connectedpapers.com\u002Fabout))_\n\n- [x] Litmaps Toggle \n\nLitmaps _([What is Litmaps?](https:\u002F\u002Fwww.litmaps.co\u002F))_\n\n- [x] scite.ai Toggle \n\nscite Smart Citations _([What are Smart Citations?](https:\u002F\u002Fwww.scite.ai\u002F))_\n\nCode, Data, Media \n\n# Code, Data and Media Associated with this Article\n\n- [x] alphaXiv Toggle \n\nalphaXiv _([What is alphaXiv?](https:\u002F\u002Falphaxiv.org\u002F))_\n\n- [x] Links to Code Toggle \n\nCatalyzeX Code Finder for Papers _([What is CatalyzeX?](https:\u002F\u002Fwww.catalyzex.com\u002F))_\n\n- [x] DagsHub Toggle \n\nDagsHub _([What is DagsHub?](https:\u002F\u002Fdagshub.com\u002F))_\n\n- [x] GotitPub Toggle \n\nGotit.pub _([What is GotitPub?](http:\u002F\u002Fgotit.pub\u002Ffaq))_\n\n- [x] Huggingface Toggle \n\nHugging Face _([What is Huggingface?](https:\u002F\u002Fhuggingface.co\u002Fhuggingface))_\n\n- [x] ScienceCast Toggle \n\nScienceCast _([What is ScienceCast?](https:\u002F\u002Fsciencecast.org\u002Fwelcome))_\n\nDemos \n\n# Demos\n\n- [x] Replicate Toggle \n\nReplicate _([What is Replicate?](https:\u002F\u002Freplicate.com\u002Fdocs\u002Farxiv\u002Fabout))_\n\n- [x] Spaces Toggle \n\nHugging Face Spaces _([What is Spaces?](https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fspaces))_\n\n- [x] Spaces Toggle \n\nTXYZ.AI _([What is TXYZ.AI?](https:\u002F\u002Ftxyz.ai\u002F))_\n\nRelated Papers \n\n# Recommenders and Search Tools\n\n- [x] Link to Influence Flower \n\nInfluence Flower _([What are Influence Flowers?](https:\u002F\u002Finfluencemap.cmlab.dev\u002F))_\n\n- [x] Core recommender toggle \n\nCORE Recommender _([What is CORE?](https:\u002F\u002Fcore.ac.uk\u002Fservices\u002Frecommender))_\n\n*   [Author](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984)\n*   [Venue](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984)\n*   [Institution](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984)\n*   [Topic](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.07984)\n\n About arXivLabs  \n\n# arXivLabs: experimental projects with community collaborators\n\narXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.\n\nBoth individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.\n\nHave an idea for a project that will add value for arXiv's community? [**Learn more about arXivLabs**](https:\u002F\u002Finfo.arxiv.org\u002Flabs\u002Findex.html).\n\n[Which authors of this paper are endorsers?](https:\u002F\u002Farxiv.org\u002Fauth\u002Fshow-endorsers\u002F2608.07984) | [Disable MathJax](javascript:setMathjaxCookie()) ([What is MathJax?](https:\u002F\u002Finfo.arxiv.org\u002Fhelp\u002Fmathjax.html)) \n\n We gratefully acknowledge support from our **major funders**, [**member institutions**](https:\u002F\u002Finfo.arxiv.org\u002Fabout\u002Fourmembers.html), , and all contributors. \n\n[About](https:\u002F\u002Finfo.arxiv.org\u002Fabout)·[Help](https:\u002F\u002Finfo.arxiv.org\u002Fhelp)·[Contact](https:\u002F\u002Finfo.arxiv.org\u002Fhelp\u002Fcontact.html)·[Subscribe](https:\u002F\u002Finfo.arxiv.org\u002Fhelp\u002Fsubscribe)·[Copyright](https:\u002F\u002Finfo.arxiv.org\u002Fhelp\u002Flicense\u002Findex.html)·[Privacy](https:\u002F\u002Finfo.arxiv.org\u002Fhelp\u002Fpolicies\u002Fprivacy_policy.html)·[Accessibility](https:\u002F\u002Finfo.arxiv.org\u002Fhelp\u002Fweb_accessibility.html)·[Operational Status (opens in new tab)](https:\u002F\u002Fstatus.arxiv.org\u002F)\n\nMajor funding support from\n\n[![Image 5: Simons Foundation](https:\u002F\u002Farxiv.org\u002Fstatic\u002Fbase\u002F1.0.1\u002Fimages\u002Ffunders\u002Fsimons-foundation.png)](https:\u002F\u002Fwww.simonsfoundation.org\u002F)[![Image 6: Simons Foundation International](https:\u002F\u002Farxiv.org\u002Fstatic\u002Fbase\u002F1.0.1\u002Fimages\u002Ffunders\u002Fsimons-foundation-international.png)](https:\u002F\u002Fwww.sfi.org.bm\u002F)[![Image 7: Schmidt Sciences](https:\u002F\u002Farxiv.org\u002Fstatic\u002Fbase\u002F1.0.1\u002Fimages\u002Ffunders\u002Fschmidt-sciences.png)](https:\u002F\u002Fwww.schmidtsciences.org\u002F)","arXiv 2608.07984 2026-08-08","2026-08-07T23:00:00Z",80,{"impact":17,"substance":18,"depth":114,"authority":19,"freshness":20,"relevant":21,"comment":259},"农业视觉大模型持续预训练新数据集与方法，发表于ECCV研讨会，对智慧农业有实质推动。",[261],{"name":255,"url":252},[263,26,121,264,265,266],"计算机视觉","持续预训练","数据集","AgriField-40K",{"code":129,"slug":130,"name":131,"path":132,"parent_code":133},[],{"status":74,"locked":14,"version":34,"method":35,"evidence":270},{"candidates":271,"agricultural_evidence":79},[272,276],{"code":129,"score":78,"supported":79,"title":273,"abstract":274,"tags":275},[],[121],[121],{"code":84,"score":21,"supported":14,"title":277,"abstract":278,"tags":279},[],[],[263],[281,282],"持续预训练 计算机视觉 精准农业 数据集","持续预训练 计算机视觉","持续预训练计算机视觉精准农业数据集-415","2026-08-11T23:57:03.188478Z",{"id":286,"title":287,"url":288,"summary":289,"summary_zh":8,"content":290,"source_name":291,"source_url":8,"published_at":292,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":293,"score_detail":294,"sources":297,"tags":299,"primary_topic":8,"secondary_topics":305,"classification":306,"search_phrases":309,"slug":312,"view_count":19,"doi":8,"paper":8,"created_at":313},414,"GROW: One-Time Training for All Grains: Open-Set Grain Recognition and Quantitative Analysis","https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.09345","Qihe Su、Mengyu Sun、Yuxi Ke、Zhuoyan Jiang、Wanneng Yang、Chenglong Huang、Ziyuan Yang于arXiv(2026-08-10)发表预印本。GROW是面向开放集谷物识别与定量分析的统一框架,无需重新训练即可识别新品种并完成定量分析。方法上,GROW首先进行类无关谷物定位,将混合谷物图像转换为独立实例用于品种计数和表型测量,然后将视觉嵌入与形态描述符融合为谷物描述符存储在可扩展的GrainBank中,查询谷物通过秩相似度加权top-k检索识别,新品种通过追加描述符纳入而无需更新部署模型。在渐进式品种扩展、不同谷物密度、背景域偏移场景下展现良好的可扩展性、鲁棒性和适应性,与联合重训练相比,GROW将平均类别注册时间从4153秒降至仅39秒,同时保持有竞争力的识别性能。","[View PDF](https:\u002F\u002Farxiv.org\u002Fpdf\u002F2608.09345)[HTML (experimental)](https:\u002F\u002Farxiv.org\u002Fhtml\u002F2608.09345v1)\n\n> Abstract:Advances in crop breeding have introduced an increasing number of grain varieties, creating a growing demand for efficient variety recognition and quantitative analysis. However, existing methods are typically trained on a fixed variety set, and incorporating newly introduced varieties requires additional data collection and model retraining. To address this limitation, we propose GROW, a framework for Grain Recognition and quantitative analysis in Open sets Without retraining. GROW first performs class-agnostic grain localization, converting mixed-grain images into individual instances for variety-wise counting and phenotypic measurement. It then combines visual embeddings and morphological descriptors into fused grain descriptors stored in an extensible GrainBank. Query grains are recognized through rank-similarity weighted top-k retrieval, and newly introduced varieties are incorporated by appending their descriptors without updating the deployed models. Extensive experiments under progressive variety expansion, varying grain densities, and background domain shifts demonstrate the scalability, robustness, and adaptability of GROW. Compared with joint retraining, GROW reduced the average category-registration time from 4153 s to only 39 s while maintaining competitive recognition performance. These results demonstrate that GROW provides an efficient and maintainable solution for extensible grain recognition, counting, and phenotypic analysis without repeated model retraining.\n\nComments:15 pages, 15 figures\nSubjects:Computer Vision and Pattern Recognition (cs.CV)\nCite as:[arXiv:2608.09345](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.09345) [cs.CV]\n(or [arXiv:2608.09345v1](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.09345v1) [cs.CV] for this version)\n[https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.09345](https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.09345)\n\narXiv-issued DOI via DataCite (pending registration)\n\n## Submission history\n\nFrom: Qihe Su [[view email](https:\u002F\u002Farxiv.org\u002Fshow-email\u002Fd3ebe409\u002F2608.09345)] \n\n**[v1]** Mon, 10 Aug 2026 09:21:51 UTC (42,162 KB)","arXiv 2608.09345 2026-08-10","2026-08-10T01:00:00Z",74,{"impact":295,"substance":114,"depth":17,"authority":19,"freshness":139,"relevant":21,"comment":296},15,"提出无需重训练的开放集谷物识别框架，显著降低品种注册时间，对农业信息化有实质贡献。",[298],{"name":291,"url":288},[26,300,301,302,303,304],"谷物识别","开放集","视觉嵌入","表型分析","GROW",[],{"status":33,"locked":14,"version":34,"method":35,"evidence":307},{"candidates":308,"agricultural_evidence":14,"reason":38},[],[310,311],"表型分析 视觉嵌入 谷物识别 开放集","表型分析 视觉嵌入","表型分析视觉嵌入谷物识别开放集-414","2026-08-11T23:57:03.167448Z"]