[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-3284":3,"related-3284":54},{"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":23,"tags":25,"search_phrases":31,"slug":34,"view_count":35,"doi":36,"paper":37,"created_at":53},3284,"Computer vision-based size estimation module for selective harvesting of lettuce","https:\u002F\u002Fdoi.org\u002F10.14719\u002Fpst.14546","Selective harvesting of leafy vegetables remains a major challenge in agricultural automation, particularly for lettuce that meet predefined size or maturity criteria. Vision-based perception systems play a critical role in enabling selective harvesting by identifying plants that meet predefined size or maturity criteria. This study presents the development and validation of computer vision-based size estimation software designed to support selective harvesting decisions in leafy vegetable production. A C #-based image processing application was developed to perform red, green and blue (RGB) colour segmentation, contour detection and pixel-to-metric size estimation using images acquired from a standard Universal Serial Bus (USB) webcam. The system converts pixel measurements into real-world dimensions through calibrated scaling factors that account for variations in camera-to-plant distance. Field experiments were conducted using three commercial lettuce cultivars Romaine, Iceberg and Curly. Vision-based canopy width measurements showed strong agreement with manual measurements (N = 60) and exhibited very strong correlations with plant weight (Romaine type: r = 0.9803; Iceberg type: r = 0.9935; Curly type: r = 0.9707; p \u003C 0.001). Linear regression models achieved high predictive performance (R² = 0.942–0.987), while error metrics indicated low prediction errors (Mean absolute error (MAE): 5.62–8.64 g; root mean square error (RMSE): 7.30–11.06 g). Bland-Altman analysis further confirmed strong agreement between vision-based estimates and manual measurements. This study primarily validates a lightweight vision-based measurement algorithm intended for future integration into robotic harvesting systems. The results demonstrate that simple RGB-based image processing can provide reliable size estimation for selective harvesting decisions in leafy vegetable production systems.","叶菜类蔬菜的选择性收获仍是农业自动化领域的一项重大挑战，尤其是对符合预定大小或成熟度标准的生菜而言。基于视觉的感知系统通过识别符合预定大小或成熟度标准的植株，在实现选择性收获中发挥着关键作用。本研究介绍了一款基于计算机视觉的大小估算软件的开发与验证，该软件旨在为叶菜类蔬菜生产中的选择性收获决策提供支持。研究开发了一款基于C#的图像处理应用程序，利用标准通用串行总线（USB）网络摄像头采集的图像，执行红绿蓝（RGB）颜色分割、轮廓检测以及像素到实际尺寸的估算。该系统通过经过校准的缩放因子将像素测量值转换为实际尺寸，并考虑了摄像头与植株之间距离的变化。田间试验采用三个商业化生菜品种——罗马生菜、结球生菜和皱叶生菜。基于视觉的冠幅测量结果与人工测量结果高度一致（N = 60），并与植株重量呈现极强的相关性（罗马型：r = 0.9803；结球型：r = 0.9935；皱叶型：r = 0.9707；p \u003C 0.001）。线性回归模型取得了较高的预测性能（R² = 0.942–0.987），误差指标显示预测误差较低（平均绝对误差（MAE）：5.62–8.64 g；均方根误差（RMSE）：7.30–11.06 g）。Bland-Altman分析进一步证实了基于视觉的估算值与人工测量值之间的高度一致性。本研究主要验证了一种轻量级视觉测量算法，旨在未来集成到机器人收获系统中。结果表明，简单的基于RGB的图像处理能够为叶菜类蔬菜生产系统中的选择性收获决策提供可靠的大小估算。",null,"Plant Science Today","2026-09-22T00:00:00Z","论文",10,false,69,{"impact":17,"substance":18,"depth":19,"authority":17,"freshness":20,"relevant":21,"comment":22},12,20,16,9,1,"该研究开发了基于RGB图像的轻量级生菜尺寸估计模块，与人工测量高度一致，为机器人选择性采收提供了可靠感知方案，具有较高的技术参考价值。",[24],{"name":10,"url":6},[26,27,28,29,30],"智慧农业","农业人工智能","生菜","机器视觉","选择性采收",[32,33],"生菜 选择性采收 机器视觉","计算机视觉 生菜 尺寸估计","生菜选择性采收机器视觉-3284",0,"10.14719\u002Fpst.14546",{"doi":36,"openalex_id":38,"authors":39,"venue":10,"cited_by_count":35,"oa_url":6,"card":46,"direction":50,"ingested_from":52},"W7214000589",[40,43],{"name":41,"orcid":42},"Erhan Kahya","https:\u002F\u002Forcid.org\u002F0000-0001-7768-9190",{"name":44,"orcid":45},"Fatma Funda Özdüven","https:\u002F\u002Forcid.org\u002F0000-0003-4286-8943",{"tldr":47,"method":48,"finding":49,"direction":50,"opportunity":51},"开发基于RGB图像处理的生菜尺寸估计软件，用于选择性采收决策。","C#图像处理，USB摄像头采集，RGB分割与轮廓检测，标定像素-实际尺寸转换。","视觉冠幅与重量强相关（r=0.97-0.99），回归模型R²达0.94-0.99，误差低。","农业人工智能与决策模型","可集成到机器人采收系统，并探索多品种、多光照条件下的鲁棒性及实时边缘部署。","openalex","2026-09-23T23:30:38.203393Z",{"total":55,"page":21,"page_size":55,"items":56},6,[57,102,130,183,220,264],{"id":58,"title":59,"url":60,"summary":61,"summary_zh":62,"content":9,"source_name":63,"source_url":60,"published_at":11,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":64,"score_detail":65,"sources":71,"tags":73,"search_phrases":76,"slug":79,"view_count":35,"doi":80,"paper":81,"created_at":101},3257,"Blueberry flow and mass estimation from harvester conveyor videos via foundation-model-assisted labeling and vision-based sensing","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.compag.2026.112432","Despite rising labor costs and growing labor shortages, adoption of machine harvesting in fresh market blueberry production has been limited by issues such as high rates of bruising and the significant loss of fruits on the ground. Machine settings strongly affect fruit quality and harvesting performance, yet they are manually configured based on operator experience, with no real-time performance feedback. Towards addressing these limitations, this paper presents a sensing proof-of-concept for analyzing berry flow through the harvester by estimating the pixel-wise proportion of blue vs green berries and volumetric flow rate from overhead RGB-D images of the harvester’s conveyor. An automated labeling pipeline combining a domain-specific YOLOv8 patch detector with SAM2-based mask generation was developed to generate a realistic large-scale berry dataset from unlabeled images captured during commercial over-the-row blueberry harvesting operations. Compared to a baseline SegFormer model trained on a smaller dataset of manually annotated images, our fine-tuned SegFormer – trained and validated on the large-scale automatically generated dataset – achieves superior performance, with green berry IoU improving from 0.377 to 0.817 and recall increasing from 0.414 to 0.911. A scene-level YOLOv8-l detection model trained and evaluated on the same dataset achieves mAP(50-95) of 0.966 and 0.898 for blue and green berries, respectively. Both models’ performance is further confirmed by an independent evaluation on a manually annotated subset of the validation images. We also compared two approaches for predicting harvest yields across five machine-setting trials: a depth-based method using the RGB-D sensor and a detection-based method using only RGB images, achieving R 2 = 0.989 and R 2 = 0.894 , respectively, against ground-truth load cell measurements from these trials. To support future research, we publicly release our datasets and models.","尽管劳动力成本上升且劳动力短缺日益加剧，但鲜食蓝莓生产中机器采收的采用仍受到果实瘀伤率高和落地果实损失严重等问题的限制。机器设置对果实品质和采收性能有很大影响，但目前仍依赖操作人员经验进行手动配置，且没有实时性能反馈。针对这些局限，本文提出了一种传感概念验证方法，通过估计采收机传送带上方RGB-D图像中蓝莓与绿果的逐像素比例以及体积流量，来分析通过采收机的果实流。我们开发了一种自动标注流程，将领域特定的YOLOv8小块检测器与基于SAM2的掩膜生成相结合，从商业跨行蓝莓采收作业期间采集的无标注图像中生成逼真的大规模蓝莓数据集。与在较小规模人工标注图像数据集上训练的基线SegFormer模型相比，我们经过微调的SegFormer——在大规模自动生成数据集上训练和验证——取得了更优性能，绿果IoU从0.377提升至0.817，召回率从0.414提升至0.911。在同一数据集上训练和评估的场景级YOLOv8-l检测模型，对蓝莓和绿果分别达到0.966和0.898的mAP(50-95)。两个模型的性能还通过在验证图像人工标注子集上的独立评估得到进一步确认。我们还比较了在五次机器设置试验中预测采收产量的两种方法：使用RGB-D传感器的基于深度的方法和使用仅RGB图像的基于检测的方法，相对于这些试验中来自称重传感器的真实测量值，分别达到R² = 0.989和R² = 0.894。为支持未来研究，我们公开了数据集和模型。","Computers and Electronics in Agriculture",83,{"impact":66,"substance":67,"depth":68,"authority":69,"freshness":20,"relevant":21,"comment":70},18,23,19,14,"该研究提出基于基础模型辅助标注与视觉传感的蓝莓流量与产量估计方法，方法新颖、数据规模大且公开数据集与模型，对智慧农业与农业机器人领域有较高参考价值。",[72],{"name":63,"url":60},[26,27,74,29,75],"农业机器人","蓝莓采收",[77,78],"蓝莓采收 机器视觉 产量估计","蓝莓收获机 传送带 视觉检测","蓝莓采收机器视觉产量估计-3257","10.1016\u002Fj.compag.2026.112432",{"doi":80,"openalex_id":82,"authors":83,"venue":63,"cited_by_count":35,"oa_url":60,"card":96,"direction":50,"ingested_from":52},"W7214013690",[84,86,88,90,93],{"name":85,"orcid":9},"A. M. Aahad",{"name":87,"orcid":9},"Pico Sankari",{"name":89,"orcid":9},"Wei Q. Yang",{"name":91,"orcid":92},"Siniša Todorović","https:\u002F\u002Forcid.org\u002F0000-0001-5793-5921",{"name":94,"orcid":95},"Joseph R. Davidson","https:\u002F\u002Forcid.org\u002F0000-0003-4388-2210",{"tldr":97,"method":98,"finding":99,"direction":50,"opportunity":100},"用收割机传送带RGB-D视频估计蓝莓流量与质量，并公开数据集和模型。","YOLOv8+SAM2自动标注，SegFormer分割，RGB-D深度与检测法估","自动标注使绿果IoU从0.377升至0.817，深度法估产R²达0.989。","可延伸至收割机参数实时闭环调控，用视觉反馈优化机器设置以减损提质。","2026-09-23T23:30:01.713624Z",{"id":103,"title":104,"url":105,"summary":106,"summary_zh":9,"content":9,"source_name":107,"source_url":9,"published_at":11,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":108,"score_detail":109,"sources":112,"tags":114,"search_phrases":117,"slug":120,"view_count":35,"doi":9,"paper":121,"created_at":129},3249,"设施番茄采摘机器人识别定位与采摘方法研究——江苏大学雷志龙等 基于改进YOLO v5-HSV融合算法识别准确率95.01%","http:\u002F\u002Fwww.qikanzj.com\u002Fhek\u002Fnyjxxb\u002Fmulu\u002F455758.html","智慧农业(中英文)期刊发表江苏大学雷志龙、刘畅、王权团队研究：针对设施单果番茄采摘需求，设计一款设施番茄智能采摘平台，主要由升降机构、采摘机构、识别与定位系统等部分组成。平台整体结构由低压一体化伺服滚珠丝杠副升降机构、六轴协作机械臂和力控末端执行器组成。基于改进YOLO v5-HSV融合算法来识别检测，通过对H分量进行图像阈值分割，提高对成熟目标果实识别的准确率，有效排除未成熟番茄和枝叶背景的干扰；通过眼在手外的标定方法，使用ZED双目相机进行定位。搭建的设施番茄采摘机样机平台在现场采摘试验中识别准确率达到95.01%，采摘成功率为87.96%，单果平均采摘时间为14.56s。","智慧农业(中英文)·江苏大学",82,{"impact":66,"substance":110,"depth":66,"authority":69,"freshness":13,"relevant":21,"comment":111},22,"核心期刊论文，方法有改进、数据完整，识别准确率与采摘成功率等指标明确，对设施农业智能装备研发有参考价值。",[113],{"name":107,"url":105},[26,27,115,29,116],"采摘机器人","设施番茄",[118,119],"江苏大学 设施番茄 采摘机器人","YOLO v5 番茄 识别定位","江苏大学设施番茄采摘机器人-3249",{"doi":9,"openalex_id":9,"authors":122,"venue":9,"cited_by_count":35,"oa_url":9,"card":123,"direction":50,"ingested_from":128},[],{"tldr":124,"method":125,"finding":126,"direction":50,"opportunity":127},"设计设施番茄智能采摘平台，融合改进YOLO v5与HSV实现识别定位与采摘。","改进YOLO v5-HSV融合算法、ZED双目相机眼在手外标定、六轴机械臂力控末","识别准确率95.01%，采摘成功率87.96%，单果平均采摘时间14.56秒。","可探索多果簇、遮挡与弱光环境下识别定位鲁棒性，并优化采摘效率与末端力控。","agent","2026-09-23T00:04:33.401768Z",{"id":131,"title":132,"url":133,"summary":134,"summary_zh":135,"content":9,"source_name":136,"source_url":133,"published_at":137,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":138,"score_detail":139,"sources":143,"tags":145,"search_phrases":148,"slug":151,"view_count":35,"doi":152,"paper":153,"created_at":182},2801,"A comprehensive survey on machine vision applications in precision agriculture: current trends and future perspectives","https:\u002F\u002Fdoi.org\u002F10.1007\u002Fs41060-026-01278-4","A comprehensive survey on machine vision applications in precision agriculture: current trends and future perspectives。International Journal of Data Science and Analytics","精准农业中机器视觉应用的综合综述：当前趋势与未来展望。《国际数据科学与分析杂志》","International Journal of Data Science and Analytics","2026-09-17T00:00:00Z",77,{"impact":66,"substance":18,"depth":140,"authority":141,"freshness":20,"relevant":21,"comment":142},17,13,"核心期刊发表的机器视觉精准农业综述，方法梳理与趋势判断具参考价值，但属综述类论文，产业影响有限。",[144],{"name":136,"url":133},[26,27,146,147,29],"精准农业","遥感监测",[149,150],"农业人工智能 智慧农业 机器视觉 精准农业","农业人工智能 智慧农业","农业人工智能智慧农业机器视觉精准农业-2801","10.1007\u002Fs41060-026-01278-4",{"doi":152,"openalex_id":154,"authors":155,"venue":136,"cited_by_count":35,"oa_url":9,"card":177,"direction":50,"ingested_from":52},"W7213471057",[156,158,160,162,165,167,169,172,175],{"name":157,"orcid":9},"Shirun Gu",{"name":159,"orcid":9},"Xinyuan Fan",{"name":161,"orcid":9},"Lihui Zhu",{"name":163,"orcid":164},"Caixia Song","https:\u002F\u002Forcid.org\u002F0000-0003-3897-7629",{"name":166,"orcid":9},"Lei Mu",{"name":168,"orcid":9},"Zichen Zhang",{"name":170,"orcid":171},"Rui Zhang","https:\u002F\u002Forcid.org\u002F0000-0002-8634-3519",{"name":173,"orcid":174},"Tong Xu","https:\u002F\u002Forcid.org\u002F0000-0001-5564-192X",{"name":176,"orcid":9},"Zhiyuan Zhang",{"tldr":178,"method":179,"finding":180,"direction":50,"opportunity":181},"综述机器视觉在精准农业中的应用现状与未来趋势。","文献综述，梳理机器视觉在精准农业中的技术路线。","机器视觉已广泛用于作物监测、病虫害识别等，但落地仍受数据与算力限制。","可聚焦轻量化模型与边缘部署，解决田间实时性与数据稀缺问题。","2026-09-17T23:30:54.103781Z",{"id":184,"title":185,"url":186,"summary":187,"summary_zh":9,"content":9,"source_name":188,"source_url":186,"published_at":189,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":190,"score_detail":191,"sources":193,"tags":195,"search_phrases":198,"slug":200,"view_count":35,"doi":201,"paper":202,"created_at":219},2620,"A deep learning-enhanced vision system for precision target spraying in chinese cabbage cultivation","https:\u002F\u002Fdoi.org\u002F10.1007\u002Fs11119-026-10444-4","A deep learning-enhanced vision system for precision target spraying in chinese cabbage cultivation。Precision Agriculture","Precision Agriculture","2026-09-16T00:00:00Z",70,{"impact":17,"substance":66,"depth":19,"authority":69,"freshness":13,"relevant":21,"comment":192},"核心期刊论文，方法新颖且时效性强，但属细分作物技术进展，产业影响范围有限。",[194],{"name":188,"url":186},[26,27,196,29,197],"精准施药","大白菜",[199,150],"农业人工智能 智慧农业 机器视觉 精准施药","农业人工智能智慧农业机器视觉精准施药-2620","10.1007\u002Fs11119-026-10444-4",{"doi":201,"openalex_id":203,"authors":204,"venue":188,"cited_by_count":35,"oa_url":9,"card":9,"direction":9,"ingested_from":52},"W7213275445",[205,207,209,211,214,217],{"name":206,"orcid":9},"Changxi Liu",{"name":208,"orcid":9},"Hang Shi",{"name":210,"orcid":9},"Hao Sun",{"name":212,"orcid":213},"Hui Zhang","https:\u002F\u002Forcid.org\u002F0000-0001-8843-7298",{"name":215,"orcid":216},"Qingda Li","https:\u002F\u002Forcid.org\u002F0000-0002-9046-2094",{"name":218,"orcid":9},"Jun Hu","2026-09-16T23:30:03.326664Z",{"id":221,"title":222,"url":223,"summary":224,"summary_zh":225,"content":9,"source_name":226,"source_url":223,"published_at":227,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":228,"score_detail":229,"sources":231,"tags":233,"search_phrases":235,"slug":238,"view_count":35,"doi":239,"paper":240,"created_at":263},2506,"LitchiInst: Instance segmentation of the main fruit-bearing branch via fruit-branch association for robotic litchi harvesting","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.biosystemseng.2026.104586","Accurate picking point localisation is a fundamental challenge in robotic harvesting. For litchi, this requires precise identification of the main fruit-bearing branch (MFBB). Current methods, however, rely on implicit structural association, failing to learn the explicit spatial correlation between fruits and branches, which leads to frequent misidentification. To address this, LitchiInst is proposed, a real-time instance segmentation framework that explicitly models this fruit-branch dependency. The core of LitchiInst is a Structure Association Guidance (SAG) mechanism, which employs SAG queries and a dedicated SAG loss to enforce spatial correlation and enable structure-aware discrimination. The framework is further enhanced by an Enhanced Feature Pyramid Network (EFPN) to capture fine-grained MFBB features and a Mask-to-3D head to convert segmentation masks into stable 3D picking points for robotic execution. LitchiInst improves MFBB average precision by 31.06% relative to the baseline while maintaining real-time processing. In real-robot experiments across three testing settings, LitchiInst achieved a 92.0% MFBB detection success rate. These results further demonstrate its practical potential as a visual guidance module for automated litchi harvesting.","准确的采摘点定位是机器人采收面临的一项基本挑战。对于荔枝而言，这需要精确识别主结果枝（MFBB）。然而，现有方法依赖隐式结构关联，无法学习果实与枝条之间显式的空间相关性，导致误识别频发。为解决这一问题，提出了LitchiInst，一种实时实例分割框架，能够显式建模这种果枝依赖关系。LitchiInst的核心是结构关联引导（SAG）机制，该机制采用SAG查询和专用的SAG损失来强化空间相关性并实现结构感知判别。该框架还通过增强特征金字塔网络（EFPN）捕获细粒度MFBB特征，并利用Mask-to-3D头将分割掩码转换为稳定的3D采摘点以供机器人执行。相较于基线，LitchiInst将MFBB平均精度提升了31.06%，同时保持实时处理能力。在三种测试设置下的真实机器人实验中，LitchiInst实现了92.0%的MFBB检测成功率。这些结果进一步证明了其作为自动化荔枝采收视觉引导模块的实际潜力。","Biosystems Engineering","2026-09-14T00:00:00Z",81,{"impact":66,"substance":110,"depth":66,"authority":69,"freshness":20,"relevant":21,"comment":230},"提出显式建模果枝空间关联的实时实例分割框架，主结果枝识别精度提升31.06%、真机成功率92%，对荔枝采摘机器人视觉引导有实质推进，但属细分作物技术进展，未达产业级突破。",[232],{"name":226,"url":223},[26,27,115,234,29],"荔枝",[236,237],"农业人工智能 采摘机器人 智慧农业 机器视觉","农业人工智能 采摘机器人","农业人工智能采摘机器人智慧农业机器视觉-2506","10.1016\u002Fj.biosystemseng.2026.104586",{"doi":239,"openalex_id":241,"authors":242,"venue":226,"cited_by_count":35,"oa_url":223,"card":258,"direction":50,"ingested_from":52},"W7213074904",[243,246,248,251,253,255],{"name":244,"orcid":245},"Yukun Qian","https:\u002F\u002Forcid.org\u002F0000-0001-9095-4024",{"name":247,"orcid":9},"Wenchang Chai",{"name":249,"orcid":250},"Haitao Wang","https:\u002F\u002Forcid.org\u002F0000-0002-8394-6410",{"name":252,"orcid":9},"Zhiyang Mai",{"name":254,"orcid":9},"Liangliang Zhou",{"name":256,"orcid":257},"Hejun Wu","https:\u002F\u002Forcid.org\u002F0000-0001-9758-5698",{"tldr":259,"method":260,"finding":261,"direction":50,"opportunity":262},"提出LitchiInst实例分割框架，通过果实-枝条关联实现荔枝主结果枝实时分割与采摘点定位。","结构关联引导机制、增强特征金字塔网络和Mask-to-3D头，基于荔枝图像数据。","主结果枝平均精度提升31.06%，真实机器人实验检测成功率达92.0%。","可探索果实-枝条显式关联机制在其他果园采摘中的迁移，并融合时序信息提升遮挡下的鲁棒性。","2026-09-15T23:30:05.541830Z",{"id":265,"title":266,"url":267,"summary":268,"summary_zh":269,"content":9,"source_name":270,"source_url":267,"published_at":271,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":272,"score_detail":273,"sources":276,"tags":278,"search_phrases":281,"slug":283,"view_count":35,"doi":284,"paper":285,"created_at":315},2288,"Field-based deep learning classification of cotton and weeds for machine-vision-assisted intra-row weed management","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.atech.2026.102567","Timely and precise weed management is essential for reducing crop-weed competition, labour requirements, and unnecessary weed-control inputs in cotton production. Reliable discrimination between cotton plants and weeds is a prerequisite for automated intra-row weed management. This study evaluated two pretrained convolutional neural network models, MobileNetV2 and DenseNet169, for binary classification of cotton and weeds using field-acquired RGB images. Images were collected from a farmer-managed cotton field in Haryana and an experimental cotton field at ICAR-Indian Agricultural Research Institute, New Delhi, during May-September 2025 under natural field conditions. The dataset comprised 2,195 weed images representing six predominant weed species and 1,211 cotton images. Images were classified into two operational classes, cotton and weed. Training images were augmented to address the original class imbalance, whereas validation and test images were retained without augmentation. Transfer learning was used to fine-tune both models, and performance was evaluated using accuracy, precision, recall, F1-score, confusion matrices, and inference time. Both models achieved high classification performance. DenseNet169 attained a test accuracy of 99.47%, compared with 98.27% for MobileNetV2. The results indicate that pretrained CNNs can provide accurate image-level discrimination between cotton and weed vegetation under the investigated field conditions. DenseNet169 therefore shows promise as a visual-perception component for machine-vision-assisted intra-row weed management. However, the present study is limited to image-level classification and does not demonstrate plant localization, continuous machine operation, actuator coordination, or field-scale weed removal. Further validation under independent locations, varying weed densities, illumination conditions, machine travel speeds, and complete perception-decision-actuation pipelines is required.","及时、精准的杂草管理对于减少棉花生产中的棉草竞争、劳动力需求和不必要的除草投入至关重要。可靠地区分棉花植株与杂草是实现行内自动化杂草管理的前提。本研究评估了两种预训练卷积神经网络模型——MobileNetV2和DenseNet169，利用田间采集的RGB图像对棉花与杂草进行二分类。图像于2025年5月至9月期间，在自然田间条件下，采集自哈里亚纳邦一处农民管理的棉田以及新德里ICAR-印度农业研究所的实验棉田。数据集包含2，195幅杂草图像，涵盖六种主要杂草种类，以及1，211幅棉花图像。图像被分为棉花和杂草两个操作类别。训练图像经过增强处理以解决原始类别不平衡问题，而验证和测试图像则保留未增强状态。采用迁移学习对两种模型进行微调，并使用准确率、精确率、召回率、F1分数、混淆矩阵和推理时间评估性能。两种模型均取得了较高的分类性能。DenseNet169的测试准确率达到99.47%，而MobileNetV2为98.27%。结果表明，预训练卷积神经网络能够在所研究的田间条件下实现棉花与杂草植被在图像层面的准确区分。因此，DenseNet169有望作为机器视觉辅助行内杂草管理的视觉感知组件。然而，本研究仅限于图像层面的分类，并未展示植株定位、机器连续作业、执行器协调或田间规模除草。仍需在独立地点、不同杂草密度、光照条件、机器行进速度以及完整的感知-决策-执行流程下进行进一步验证。","Smart Agricultural Technology","2026-09-11T00:00:00Z",67,{"impact":17,"substance":66,"depth":19,"authority":141,"freshness":274,"relevant":21,"comment":275},8,"基于田间RGB图像的棉花与杂草深度学习分类研究，DenseNet169测试准确率达99.47%，方法扎实但仅限图像级分类，尚缺定位与执行环节验证，属细分领域技术进展。",[277],{"name":270,"url":267},[26,27,279,280,29],"棉花","杂草识别",[282,150],"农业人工智能 智慧农业 机器视觉 杂草识别","农业人工智能智慧农业机器视觉杂草识别-2288","10.1016\u002Fj.atech.2026.102567",{"doi":284,"openalex_id":286,"authors":287,"venue":270,"cited_by_count":35,"oa_url":267,"card":310,"direction":50,"ingested_from":52},"W7212315230",[288,290,293,296,299,301,304,306,308],{"name":289,"orcid":9},"Shaik Nasreen",{"name":291,"orcid":292},"Roaf Ahmad Parray","https:\u002F\u002Forcid.org\u002F0000-0002-8303-1990",{"name":294,"orcid":295},"Parveen Dhanger","https:\u002F\u002Forcid.org\u002F0000-0001-5250-0275",{"name":297,"orcid":298},"Rishi Raj","https:\u002F\u002Forcid.org\u002F0009-0002-7918-1141",{"name":300,"orcid":9},"Tapan Kumar Khura",{"name":302,"orcid":303},"P. Sahoo","https:\u002F\u002Forcid.org\u002F0000-0002-3888-8506",{"name":305,"orcid":9},"Tushar Dhar",{"name":307,"orcid":9},"Prajwal R",{"name":309,"orcid":9},"Sripriyanka S. Nalla",{"tldr":311,"method":312,"finding":313,"direction":50,"opportunity":314},"用MobileNetV2和DenseNet169对田间棉花与杂草图像做二分类，验证机器视觉除草可行性","迁移学习微调两种预训练CNN，使用2025年田间RGB图像共3406张。","DenseNet169测试准确率达99.47%，优于MobileNetV2的98.27%。","可延伸至植株定位、多光照与密度条件下的感知-决策-执行全流程田间验证。","2026-09-13T23:30:04.183782Z"]