[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-5534":3,"related-5534":76},{"id":4,"title":5,"url":6,"summary":7,"summary_zh":8,"content":8,"source_name":9,"source_url":8,"published_at":10,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":14,"score_detail":15,"sources":23,"tags":25,"primary_topic":31,"secondary_topics":37,"classification":44,"search_phrases":61,"slug":64,"view_count":65,"doi":8,"paper":66,"created_at":75},5534,"DC-YOLO：应对遮挡与光照变化的湖羊行为检测模型","https:\u002F\u002Ffinance.sina.com.cn\u002Froll\u002F2026-10-19\u002Fdoc-inivumaf7730137.shtml","中国农业大学团队针对集约化养殖中多目标严重遮挡、光照多变与类别不均问题，引入自适应细节—上下文注意力机制构建模块，并采用边界框相似度软抑制防止误删遮挡目标，预训练光照编码器迁移参数应对光照变化。数据集来自湖羊养殖场全天候监控，含505张标注图像、5类行为。",null,"智慧农业（中英文）2026-10-19","2026-10-19T00:00:00Z","论文",10,false,75,{"impact":16,"substance":17,"depth":18,"authority":19,"freshness":20,"relevant":21,"comment":22},18,21,17,13,6,1,"论文提出面向湖羊行为检测的DC-YOLO模型，方法有新意、数据来自真实养殖场，属智慧养殖细分领域实质进展，值得精选。",[24],{"name":9,"url":6},[26,27,28,29,30],"农业人工智能","智慧养殖","目标检测","湖羊","动物行为检测",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},"C2","ruminants","草食牲畜","\u002Ftopic\u002Flivestock\u002Fruminants","C",[38],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},"I2","agricultural-ai","农业AI与大模型","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-ai","I",{"status":45,"locked":13,"version":46,"method":47,"evidence":48},"classified","2026-10-22.2","rules",{"candidates":49,"agricultural_evidence":52},[50,56],{"code":32,"score":51,"supported":52,"title":53,"abstract":54,"tags":55},12,true,[29],[29],[29],{"code":39,"score":12,"supported":52,"title":57,"abstract":59,"tags":60},[58],"YOLO",[],[26,28],[62,63],"中国农业大学 湖羊 行为检测","DC-YOLO 遮挡 光照","中国农业大学湖羊行为检测-5534",0,{"doi":8,"openalex_id":8,"authors":67,"venue":8,"cited_by_count":65,"oa_url":8,"card":68,"direction":72,"ingested_from":74},[],{"tldr":69,"method":70,"finding":71,"direction":72,"opportunity":73},"提出DC-YOLO模型，用于湖羊行为检测，解决遮挡与光照变化问题。","自适应细节-上下文注意力、边界框相似度软抑制、光照编码器迁移，505张标注图像。","模型在遮挡和光照变化下有效检测湖羊行为，减少误删。","智慧农业 \u002F 农业物联网","可探索多模态融合与轻量化部署，提升复杂养殖场景下的实时行为监测能力。","agent","2026-10-22T00:07:09.998247Z",{"total":20,"page":21,"page_size":20,"items":77},[78,148,229,278,374,438],{"id":79,"title":80,"url":81,"summary":82,"summary_zh":83,"content":8,"source_name":84,"source_url":81,"published_at":85,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":86,"score_detail":87,"sources":92,"tags":94,"primary_topic":98,"secondary_topics":99,"classification":101,"search_phrases":119,"slug":122,"view_count":21,"doi":123,"paper":124,"created_at":147},4975,"Detection of mounting behavior and mounted ewes in group-housed environments using HSM-YOLO algorithm","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.atech.2026.102639","The detection of mounting behavior and the mounted ewe is critical for estrus monitoring in sheep. However, existing methods suffer from low detection efficiency and high false detection rates due to the similarity between non-sexual and sexual mounting behaviors and the complexity of group-housed environments. To address these challenges, this study introduced a fine-grained classification of mounting behaviors into sexual and non-sexual categories, and further proposed an improved detection algorithm named Hu Sheep Mounting-YOLO (HSM-YOLO) based on the YOLOv11n architecture for detecting sheep mounting behavior and the corresponding mounted ewes. The HSM-YOLO algorithm employed depthwise separable convolution (DWSConv) to reduce the model complexity. To improve behavioral feature extraction and pose discrimination capabilities, it further integrated the C3k2_RepViTBlock-RepBottleneck (C3k2_BiRep) module, an efficient multi-scale attention (EMA) module, and the Inner-SIoU loss function. To evaluate the performance of the proposed algorithm, comparative experiments were conducted on a multiple scenarios dataset. The experimental results showed that the HSM-YOLO model achieved an F1 score of 88.1%, a mean Average Precision (mAP) of 90.2%, with 2.2M parameters and 5.4G floating-point operations (FLOPs). Compared to the baseline YOLOv11n model, HSM-YOLO improved the F1 score by 5.8% and the mAP by 5.7%, while reducing both model parameters and computational complexity. When deployed and tested on an RTX 4090, the system delivered an average detection speed of 47.9 frames per second (FPS). Moreover, despite considerable variation in detection performance across the four different imaging scenarios, HSM-YOLO consistently demonstrated superior detection accuracy. These results confirm that the HSM-YOLO provides a robust and efficient solution for detecting sheep mounting behaviors and mounted ewes in complex and variable scenarios, paving the way for the development of automated estrus monitoring systems.","检测羊只的爬跨行为及被爬跨母羊对于绵羊发情监测至关重要。然而，由于非性爬跨行为与性爬跨行为之间的相似性以及群养环境的复杂性，现有方法存在检测效率低和误检率高的问题。为解决这些挑战，本研究将爬跨行为细粒度分类为性爬跨和非性爬跨两类，并进一步提出了一种基于YOLOv11n架构的改进检测算法——湖羊爬跨-YOLO（HSM-YOLO），用于检测绵羊爬跨行为及相应的被爬跨母羊。HSM-YOLO算法采用深度可分离卷积（DWSConv）以降低模型复杂度。为提高行为特征提取和姿态判别能力，该算法进一步集成了C3k2_RepViTBlock-RepBottleneck（C3k2_BiRep）模块、高效多尺度注意力（EMA）模块以及Inner-SIoU损失函数。为评估所提算法的性能，在多场景数据集上进行了对比实验。实验结果表明，HSM-YOLO模型的F1分数达到88.1%，平均精度均值（mAP）达到90.2%，参数量为2.2M，浮点运算量为5.4G。与基线YOLOv11n模型相比，HSM-YOLO的F1分数提高了5.8%，mAP提高了5.7%，同时降低了模型参数量和计算复杂度。在RTX 4090上部署测试时，系统平均检测速度达到47.9帧每秒（FPS）。此外，尽管在四种不同成像场景下检测性能存在较大差异，HSM-YOLO始终展现出优越的检测精度。这些结果证实，HSM-YOLO为复杂多变场景下检测绵羊爬跨行为和被爬跨母羊提供了一种鲁棒且高效的解决方案，为自动化发情监测系统的开发奠定了基础。","Smart Agricultural Technology","2026-10-12T00:00:00Z",81,{"impact":16,"substance":88,"depth":16,"authority":89,"freshness":90,"relevant":21,"comment":91},22,14,9,"提出HSM-YOLO算法实现羊只爬跨行为与受体母羊检测，为自动化发情监测提供高效方案，方法新颖、指标扎实，具行业参考价值。",[93],{"name":84,"url":81},[95,26,28,96,97],"智慧农业","畜牧养殖","羊只发情监测",{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},[100],{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},{"status":45,"locked":13,"version":46,"method":47,"evidence":102},{"candidates":103,"agricultural_evidence":52},[104,108,114],{"code":39,"score":19,"supported":52,"title":105,"abstract":106,"tags":107},[58],[58],[26,28],{"code":32,"score":90,"supported":52,"title":109,"abstract":110,"tags":113},[],[29,111,112],"绵羊","sheep",[],{"code":115,"score":21,"supported":13,"title":116,"abstract":117,"tags":118},"I1",[],[],[95],[120,121],"HSM-YOLO 羊只爬跨行为 检测","绵羊 发情监测 目标检测","HSM-YOLO羊只爬跨行为检测-4975","10.1016\u002Fj.atech.2026.102639",{"doi":123,"openalex_id":125,"authors":126,"venue":84,"cited_by_count":65,"oa_url":81,"card":140,"direction":144,"ingested_from":146},"W7221573347",[127,129,131,133,136,138],{"name":128,"orcid":8},"Yanchao Wang",{"name":130,"orcid":8},"Ming Gao",{"name":132,"orcid":8},"Zhixin Hua",{"name":134,"orcid":135},"Asim Shakeel","https:\u002F\u002Forcid.org\u002F0000-0003-2942-9081",{"name":137,"orcid":8},"Chengyu Ren",{"name":139,"orcid":8},"Kaiying Wang",{"tldr":141,"method":142,"finding":143,"direction":144,"opportunity":145},"提出HSM-YOLO算法，在群养环境下检测羊爬跨行为及被爬跨母羊，用于发情监测。","基于YOLOv11n改进，引入DWSConv、C3k2_BiRep、EMA和In","F1达88.1%、mAP达90.2%，参数量2.2M，较基线提升且更轻量，检测速度47.9 FPS。","农业人工智能与决策模型","可延伸至多物种发情行为细粒度识别与轻量化边缘部署，结合时序信息提升监测实时性。","openalex","2026-10-15T23:30:03.393878Z",{"id":149,"title":150,"url":151,"summary":152,"summary_zh":8,"content":8,"source_name":153,"source_url":151,"published_at":154,"category":11,"cover_url":8,"hotness":155,"is_selected":13,"score":156,"score_detail":157,"sources":160,"tags":164,"primary_topic":168,"secondary_topics":169,"classification":171,"search_phrases":201,"slug":204,"view_count":205,"doi":206,"paper":207,"created_at":228},2780,"基于深度学习数据增强技术的肉牛行为识别 与养殖状态评估研究","https:\u002F\u002Fdoi.org\u002F10.65436\u002Fhssj.v1i9.50","摘 要：肉牛行为的准确识别是智慧养殖中个体状态监测和精细化管理的重要基础。针对实际养殖场景中光照变化、行为姿态差异以及训练样本规模对识别性能影响不明确等问题，本文以9类肉牛典型行为图像为研究对象，构建基于深度学习骨干网络与组合数据增强的行为识别流程。数据集包含饮水、进食、争斗、舔舐、躺卧、爬跨、探究、站立和行走9类行为，覆盖低光、正常光照、高光和黑暗4种光照条件，共4500张图像。首先，在统一训练条件下比较ResNet-50、DenseNet-121、EfficientNet-B0和Swin-T四种骨干网络。实验结果表明，EfficientNet-B0的测试准确率为98.37%±1.36%，参数量为4.02 M，MACs为0.41 G；其准确率较ResNet-50高0.07个百分点，较DenseNet-121和Swin-T均高0.59个百分点，同时具有四种网络中最低的参数量与计算量，因此选其作为后续实验的固定骨干。进一步设置不同训练样本规模实验与数据增强消融实验，以分析样本量和翻转、颜色抖动、旋转、随机擦除等增强操作的影响。实验二结果表明，当每类训练样本由每类10张增至全量训练集（每类400张）时，测试准确率由69.70%±0.76%提升至98.37%±1.36%，训练样本规模对识别性能影响显著，且随样本量增大边际收益总体呈递减趋势。数据增强消融实验表明，逐项去除翻转、颜色抖动、旋转和随机擦除后，测试准确率较完整增强基线（98.37%±1.36%）分别下降2.74、3.11、2.89和2.96个百分点，验证了各增强操作对复杂光照与姿态条件下识别性能的贡献，其中颜色抖动的作用最为显著。研究可为复杂光照条件下肉牛行为的自动识别与智能养殖监测提供参考。 基金项目：安徽省教育厅优秀青年教师培育项目（YQYB2026117）；安徽省智慧农业技术与装备重点实验室开放基金（AEC2026016；AEC2025009）；安徽省教育厅自然科学研究项目（重大）（2024AH040217）；省级质量工程-教学研究项目（2023jyxm1009；2023xjzlts117；2023sdxx145）;基于人工智能下的家禽康养与疾病预警技术研究（S202413620055）；基于自监督学习与对抗增强的小样本信号的智能感知研究（2025SK019）；基于深度学习的生猪识别与康养研究（S202513620083）","人文与社会研究学报","2026-09-17T00:00:00Z",25,78,{"impact":158,"substance":88,"depth":16,"authority":51,"freshness":12,"relevant":21,"comment":159},16,"论文以4500张多光照肉牛行为图像系统比较四种骨干网络并完成样本规模与数据增强消融实验，方法扎实、结论可靠，对复杂光照下智能养殖监测有直接参考价值，值得进入每日精选。",[161,162],{"name":153,"url":151},{"name":153,"url":163},"https:\u002F\u002Fdoi.org\u002F10.65436\u002Fhssj.v1i9.48",[26,165,27,166,167],"深度学习","数据增强","肉牛行为识别",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[170],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},{"status":45,"locked":13,"version":46,"method":47,"evidence":172},{"candidates":173,"agricultural_evidence":52},[174,178,184,191,197],{"code":39,"score":19,"supported":52,"title":175,"abstract":176,"tags":177},[165],[165],[26,165],{"code":32,"score":179,"supported":52,"title":180,"abstract":182,"tags":183},11,[181],"肉牛",[181],[],{"code":185,"score":186,"supported":52,"title":187,"abstract":188,"tags":190},"C1",3,[],[189],"生猪",[],{"code":192,"score":186,"supported":52,"title":193,"abstract":194,"tags":196},"C3",[],[195],"家禽",[],{"code":115,"score":186,"supported":52,"title":198,"abstract":199,"tags":200},[],[95],[],[202,203],"农业人工智能 肉牛行为识别 数据增强 智慧养殖","农业人工智能 肉牛行为识别","农业人工智能肉牛行为识别数据增强智慧养殖-2780",2,"10.65436\u002Fhssj.v1i9.50",{"doi":206,"openalex_id":208,"authors":209,"venue":153,"cited_by_count":65,"oa_url":151,"card":223,"direction":72,"ingested_from":146},"W7213445826",[210,212,214,216,218,220],{"name":211,"orcid":8},"孙帅",{"name":213,"orcid":8},"赵永才",{"name":215,"orcid":8},"李慢",{"name":217,"orcid":8},"郜静茹",{"name":219,"orcid":8},"关曼玉",{"name":221,"orcid":222},"Yanbo Li","https:\u002F\u002Forcid.org\u002F0000-0001-5144-2309",{"tldr":224,"method":225,"finding":226,"direction":72,"opportunity":227},"比较四种骨干网络并做数据增强消融，实现9类肉牛行为识别与养殖状态评估。","4500张9类行为图像，ResNet-50、DenseNet-121、Effic","EfficientNet-B0最优（98.37%，4.02M参数）；样本量影响显著，颜色抖动贡献最大","可探索轻量模型在边缘设备实时部署及跨牧场域适应，并引入时序视频行为识别。","2026-09-17T23:30:22.897005Z",{"id":230,"title":231,"url":232,"summary":233,"summary_zh":8,"content":8,"source_name":234,"source_url":8,"published_at":235,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":236,"score_detail":237,"sources":240,"tags":242,"primary_topic":246,"secondary_topics":247,"classification":249,"search_phrases":265,"slug":268,"view_count":65,"doi":8,"paper":269,"created_at":277},5363,"牛肉牛区域特征计算机视觉分析:基于 YOLO 与深度估计的牛体况自动评分","https:\u002F\u002Fbioengineer.org\u002Fcameras-learn-to-read-a-cows-spine-tail-head-and-body-shape-to-score-its-condition","Islam M.N., Schrick F.N., Edwards J.L., Burns R.T., Qi H., Gan H. (2026) 开发一种基于区域特定计算机视觉特征的牛肉牛体况自动评分(BCS)方法。研究通过 YOLO 检测 + 深度估计提取脊柱、尾根、体廓等区域特征,使用 SHAP 进行可解释 AI 分析,并以 XGBoost 融合。在 5 折交叉验证中两种模式均稳定。验证集仅 72 只动物且极端值样本少,深度图来自推断而非实测。框架可作为后续发展的方法学基础。DOI: 10.1016\u002Fj.atech.2026.102628。","Smart Agricultural Technology Vol.15 Article 102628 2026-10-15","2026-10-15T00:00:00Z",68,{"impact":51,"substance":238,"depth":18,"authority":19,"freshness":20,"relevant":21,"comment":239},20,"方法学上有新意，但验证样本仅72头且深度图为推断值，属细分领域论文进展，未达每日精选门槛。",[241],{"name":234,"url":232},[95,26,243,244,245],"计算机视觉","肉牛养殖","体况评分",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[248],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},{"status":45,"locked":13,"version":46,"method":47,"evidence":250},{"candidates":251,"agricultural_evidence":52},[252,257,261],{"code":39,"score":253,"supported":52,"title":254,"abstract":255,"tags":256},24,[243,58],[243,58],[26,243],{"code":32,"score":51,"supported":52,"title":258,"abstract":259,"tags":260},[181],[181],[244],{"code":115,"score":21,"supported":13,"title":262,"abstract":263,"tags":264},[],[],[95],[266,267],"YOLO 深度估计 牛体况评分","牛肉牛 计算机视觉 BCS","YOLO深度估计牛体况评分-5363",{"doi":8,"openalex_id":8,"authors":270,"venue":8,"cited_by_count":65,"oa_url":8,"card":271,"direction":275,"ingested_from":74},[],{"tldr":272,"method":273,"finding":274,"direction":275,"opportunity":276},"用YOLO检测与深度估计提取牛体区域特征，实现肉牛体况自动评分。","YOLO检测、深度估计、SHAP可解释分析与XGBoost融合，5折交叉验证。","区域特征融合评分稳定，但验证集仅72头且深度图为推断值。","农业遥感与作物表型","可引入实测深度传感器与更大样本，提升极端体况评分鲁棒性。","2026-10-20T02:47:54.361149Z",{"id":279,"title":280,"url":281,"summary":282,"summary_zh":283,"content":8,"source_name":284,"source_url":281,"published_at":285,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":286,"score_detail":287,"sources":290,"tags":292,"primary_topic":296,"secondary_topics":297,"classification":299,"search_phrases":348,"slug":351,"view_count":352,"doi":353,"paper":354,"created_at":373},4422,"Artificial intelligence driven framework for enhancing sustainability and resilience in communal cattle insurance in Lupane District Zimbabwe","https:\u002F\u002Fdoi.org\u002F10.1007\u002Fs44279-026-00786-y","Communal cattle farming in Sub-Saharan Africa continues to face destabilization from non-linear climate variability, yet insurance penetration remains negligible. In Lupane District, Zimbabwe, uptake is exceptionally low at approximately 0.01%, despite recurrent droughts that threaten household livelihoods. This study aimed to develop an AI-driven framework to enhance sustainability and resilience in communal cattle insurance within the district. A positivist research philosophy guided the study, and a quantitative approach was adopted. The Kothari sample size determination formula was applied to derive a statistically representative sample of 219 households from a population of 49,841 communal cattle farmers distributed across four villages in 23 wards. Data were collected through KoboCollect, transmitted to secure servers, and analyzed using JAMOVI version 2.6.44. Descriptive statistics indicated that the majority of farmers were male (62%), with an average age of 47 years and herd sizes ranging from 3 to 25 cattle. Education levels were generally low, with 54% reporting primary schooling and only 12% having secondary or higher qualifications. Income sources were largely subsistence-based, and 71% of respondents relied on cattle as their primary asset. Binary logistic regression revealed that affordability (β = 1.25, p \u003C 0.01) and awareness (β = 0.99, p \u003C 0.01) were the strongest determinants of adoption, while institutional trust (β = 0.74, p \u003C 0.05) emerged as a significant direct predictor of adoption willingness. The study was theoretically anchored in the Technology Acceptance Model (TAM), which explains how perceived usefulness and perceived ease of use influence uptake. Informed by findings from Objectives 1 and 2, the study proposes the Tri-Modular AI Resilience Engine (TMAIE). TMAIE integrates Convolutional Neural Networks (CNN) for biometric cattle identification, Recurrent Neural Networks (RNN) for satellite-derived NDVI index triggers, and Transformer-based Natural Language Processing (NLP) localized in IsiNdebele and ChiShona for inclusive voice-based access. The study findings suggest that adoption is not a linear process but an adaptive evolution requiring both technological and social mechanisms. The proposed framework reflects these relationships through the integration of technological precision and community-based mediation. Overall, the findings provide empirical evidence on the determinants of cattle insurance adoption and inform the development of a conceptual AI-driven framework that may support future climate-resilient livestock insurance innovations, subject to further validation and testing.","撒哈拉以南非洲的社区养牛业持续面临非线性气候变率的冲击，然而保险渗透率仍然微不足道。在津巴布韦卢潘内区，尽管反复发生的干旱威胁着家庭生计，但参保率仍极低，约为0.01%。本研究旨在开发一个人工智能驱动的框架，以提升该区社区养牛保险的可持续性和韧性。本研究以实证主义研究哲学为指导，采用定量研究方法。应用Kothari样本量确定公式，从分布于23个选区4个村庄的49841名社区养牛户总体中，推导出具有统计代表性的219户样本。数据通过KoboCollect采集，传输至安全服务器，并使用JAMOVI 2.6.44版本进行分析。描述性统计表明，大多数农户为男性（62%），平均年龄47岁，牛群规模为3至25头。教育水平普遍较低，54%报告仅接受过小学教育，仅12%拥有中学或更高学历。收入来源大多以自给自足为基础，71%的受访者将牛作为其主要资产。二元逻辑回归显示，可负担性（β = 1.25，p \u003C 0.01）和认知度（β = 0.99，p \u003C 0.01）是参保的最强决定因素，而制度信任（β = 0.74，p \u003C 0.05）则是参保意愿的显著直接预测因素。本研究在理论上以技术接受模型（TAM）为锚点，该模型解释了感知有用性和感知易用性如何影响采纳行为。基于目标1和目标2的研究结果，本研究提出了三模块人工智能韧性引擎（TMAIE）。TMAIE整合了用于牛只生物特征识别的卷积神经网络（CNN）、用于卫星衍生NDVI指数触发的循环神经网络（RNN），以及基于Transformer的、以恩德贝莱语和绍纳语本地化的自然语言处理（NLP），以实现包容性的语音访问。研究结果表明，采纳并非线性过程，而是一种需要技术和社会机制共同作用的自适应演化。所提出的框架通过整合技术精确性与社区中介来反映这些关系。总体而言，研究结果提供了关于养牛保险采纳决定因素的实证证据，并为","Discover Agriculture","2026-10-06T00:00:00Z",70,{"impact":51,"substance":17,"depth":18,"authority":51,"freshness":288,"relevant":21,"comment":289},8,"以津巴布韦Lupane地区社区养牛保险为对象，提出融合CNN牛只识别、NDVI遥感触发与本地语言NLP的AI框架，方法具体、数据扎实，但属区域性案例研究，对国内农业信息化有借鉴意义而非直接影响。",[291],{"name":284,"url":281},[293,95,26,294,295],"数字乡村","农业保险","遥感监测",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[298],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},{"status":45,"locked":13,"version":46,"method":47,"evidence":300},{"candidates":301,"agricultural_evidence":52},[302,307,312,317,323,329,334,338,343],{"code":32,"score":179,"supported":52,"title":303,"abstract":305,"tags":306},[304],"cattle",[304],[],{"code":39,"score":90,"supported":52,"title":308,"abstract":310,"tags":311},[309],"artificial intelligence",[],[26],{"code":36,"score":186,"supported":52,"title":313,"abstract":314,"tags":316},[],[315],"livestock",[],{"code":318,"score":186,"supported":52,"title":319,"abstract":320,"tags":322},"H7",[],[321],"干旱",[],{"code":324,"score":186,"supported":52,"title":325,"abstract":326,"tags":328},"I7",[],[327],"研究方法",[],{"code":330,"score":21,"supported":13,"title":331,"abstract":332,"tags":333},"G3",[],[],[294],{"code":115,"score":21,"supported":13,"title":335,"abstract":336,"tags":337},[],[],[95],{"code":339,"score":21,"supported":13,"title":340,"abstract":341,"tags":342},"I3",[],[],[295],{"code":344,"score":21,"supported":13,"title":345,"abstract":346,"tags":347},"I6",[],[],[293],[349,350],"津巴布韦 Lupane 牛保险","AI 牲畜保险 框架","津巴布韦Lupane牛保险-4422",5,"10.1007\u002Fs44279-026-00786-y",{"doi":353,"openalex_id":355,"authors":356,"venue":284,"cited_by_count":65,"oa_url":366,"card":367,"direction":72,"ingested_from":146},"W7220655164",[357,360,363],{"name":358,"orcid":359},"Nyasha Nyakuchena","https:\u002F\u002Forcid.org\u002F0009-0005-3219-2275",{"name":361,"orcid":362},"Joseph P. Musara","https:\u002F\u002Forcid.org\u002F0000-0002-2226-3616",{"name":364,"orcid":365},"Emmanuel Zivenge","https:\u002F\u002Forcid.org\u002F0000-0003-4740-8225","https:\u002F\u002Flink.springer.com\u002Fcontent\u002Fpdf\u002F10.1007\u002Fs44279-026-00786-y.pdf",{"tldr":368,"method":369,"finding":370,"direction":371,"opportunity":372},"研究提出AI驱动的三模块框架，以提升津巴布韦Lupane地区社区牛保险的可持续性和韧性。","对219户农户进行问卷调查，用二元逻辑回归分析，并设计CNN、RNN和Trans","可负担性和意识是保险采纳的最强决定因素，机构信任也显著影响采纳意愿。","数字乡村与农业信息化","可验证TMAIE框架在类似地区的适用性，并探索AI语音交互对低教育水平农户保险采纳的长期影响。","2026-10-08T23:30:11.018464Z",{"id":375,"title":376,"url":377,"summary":378,"summary_zh":379,"content":8,"source_name":380,"source_url":377,"published_at":381,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":382,"score_detail":383,"sources":385,"tags":387,"primary_topic":390,"secondary_topics":391,"classification":393,"search_phrases":411,"slug":414,"view_count":21,"doi":415,"paper":416,"created_at":437},4244,"Automated dairy cattle lameness assessment using a 38-keypoint whole-body YOLOv8 pose framework","https:\u002F\u002Fdoi.org\u002F10.14202\u002Fvetworld.2026.4286-4306","Background and Aim: Lameness is a major health, welfare, and production concern in dairy cattle, whereas conventional visual locomotion scoring is intermittent and susceptible to observer-dependent variation. Markerless computer vision may enable objective and repeatable locomotion monitoring under routine farm conditions. This study aimed to develop a 38-keypoint whole-body pose-estimation framework for automated dairy cattle lameness assessment and evaluate its agreement with blinded Sprecher locomotion scoring. Materials and Methods: A You Only Look Once version 8 (YOLOv8)-pose framework was developed using 12,000 annotated images obtained from 449 Holstein cows. Thirty-eight anatomical keypoints represented the head, spine, pelvis, forelimbs, hindlimbs, and hooves. Four model and input resolution configurations were evaluated. Joint-angle time series, coefficients of variation, and whole-body asymmetry features were derived from detected keypoints. For independent clinical validation, 120 cattle videos, obtained from animals with no overlap at the animal level with the 449-cow model-development cohort, were evaluated blindly by three assessors using the five-point Sprecher scale. Inter-observer reliability, consensus scores, rank associations, group differences, and exploratory receiver operating characteristic analyses were determined. Results: YOLOv8l-pose at 2560 × 2560 pixels showed the best performance, achieving pose mean average precision at an intersection over union threshold of 0.5 of 83.11%, mean average precision across thresholds of 0.5–0.95 of 37.39%, and recall of 91.74%. All three assessors assigned identical Sprecher scores to 87.5% of videos. Pairwise quadratic-weighted Cohen κ ranged from 0.843 to 0.926, Fleiss κ was 0.779, and intraclass correlation coefficients were 0.883 and 0.958 for single and average measurements, respectively. The continuous artificial intelligence (AI) score correlated significantly with consensus Sprecher scores (Spearman ρ = 0.234, p = 0.010), and score distributions differed among consensus groups (Kruskal–Wallis H = 13.09, p = 0.0014). Exploratory discrimination of consensus Sprecher scores ≥3, using a threshold internally derived from and evaluated within the same stratified validation cohort (not externally validated), yielded an area under the curve of 0.875; at an AI threshold of 2.9, sensitivity and specificity were 62.5% and 98.2%, respectively. Conclusion: The 38-keypoint whole-body model demonstrates technical feasibility as an anatomically comprehensive, sensor-free approach for quantifying dairy cattle locomotion. Because the clinical validation cohort was deliberately stratified and diagnostic thresholds were internally derived, its agreement with independent visual assessment should be interpreted as evidence of potential decision-support utility rather than established clinical performance. Prospective external validation and independent calibration of diagnostic thresholds in an unselected cohort are warranted before clinical or commercial deployment. Keywords: artificial intelligence, bovine lameness, computer vision, dairy cattle, locomotion assessment, pose-estimation, Sprecher score, whole-body keypoints.","背景与目的：跛行是奶牛重要的健康、福利和生产问题，而传统的视觉运动评分具有间歇性且易受观察者差异影响。无标记计算机视觉可能有助于在常规农场条件下实现客观且可重复的运动监测。本研究旨在开发一个38关键点全身姿态估计框架，用于自动化奶牛跛行评估，并评估其与盲法Sprecher运动评分的一致性。材料与方法：使用来自449头荷斯坦奶牛的12,000张标注图像，开发了You Only Look Once第8版（YOLOv8）-pose框架。38个解剖关键点代表头部、脊柱、骨盆、前肢、后肢和蹄部。评估了四种模型和输入分辨率配置。从检测到的关键点中提取关节角度时间序列、变异系数和全身不对称特征。为进行独立临床验证，由三名评估者使用五分制Sprecher量表对120头牛的视频进行盲法评估，这些动物在个体层面与449头模型开发队列无重叠。确定了观察者间可靠性、共识评分、秩相关、组间差异和探索性受试者工作特征分析。结果：2560×2560像素下的YOLOv8l-pose表现最佳，在交并比阈值为0.5时姿态平均精度为83.11%，在0.5–0.95阈值范围内的平均精度为37.39%，召回率为91.74%。三名评估者对87.5%的视频给出了相同的Sprecher评分。两两二次加权Cohen κ范围为0.843至0.926，Fleiss κ为0.779，单次测量和平均测量的组内相关系数分别为0.883和0.958。连续人工智能（AI）评分与共识Sprecher评分显著相关（Spearman ρ = 0.234，p = 0.010），且各共识组之间的评分分布不同（Kruskal–Wallis H = 13.09，p = 0.0014）。使用在同一分层验证队列内部推导并评估的阈值（未经外部验证），对共识Sprecher评分≥3进行探索性判别时，曲线下面积为0.875；在AI阈值为2.9时，敏感性和特异性分别为62.5%和98.2%。结论：38关键点全身模型表现","Veterinary World","2026-10-03T00:00:00Z",80,{"impact":16,"substance":88,"depth":16,"authority":19,"freshness":90,"relevant":21,"comment":384},"基于12000张标注图像与120头独立验证视频的38关键点YOLOv8姿态估计框架，为奶牛跛行无接触客观评估提供技术可行性证据，方法新颖、数据扎实，但阈值内推、缺乏外部验证，属细分领域技术进展。",[386],{"name":380,"url":377},[95,26,243,388,389],"动物健康","奶牛养殖",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[392],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},{"status":45,"locked":13,"version":46,"method":47,"evidence":394},{"candidates":395,"agricultural_evidence":52},[396,402,407],{"code":32,"score":397,"supported":52,"title":398,"abstract":399,"tags":401},15,[304],[400,304],"奶牛",[389],{"code":39,"score":179,"supported":52,"title":403,"abstract":404,"tags":406},[],[243,309,405],"computer vision",[26,243],{"code":115,"score":21,"supported":13,"title":408,"abstract":409,"tags":410},[],[],[95],[412,413],"YOLOv8 奶牛 跛行 姿态估计","奶牛跛行 自动评估 计算机视觉","YOLOv8奶牛跛行姿态估计-4244","10.14202\u002Fvetworld.2026.4286-4306",{"doi":415,"openalex_id":417,"authors":418,"venue":380,"cited_by_count":65,"oa_url":431,"card":432,"direction":371,"ingested_from":146},"W7218963807",[419,422,425,428],{"name":420,"orcid":421},"А В Греченева","https:\u002F\u002Forcid.org\u002F0000-0002-7341-5237",{"name":423,"orcid":424},"Maxim Baknin","https:\u002F\u002Forcid.org\u002F0000-0002-3124-6252",{"name":426,"orcid":427},"Sergey Vladimirovich Akchurin","https:\u002F\u002Forcid.org\u002F0000-0002-6822-0013",{"name":429,"orcid":430},"Ирина Владимировна Акчурина","https:\u002F\u002Forcid.org\u002F0000-0002-1975-2929","https:\u002F\u002Fveterinaryworld.org\u002FVol.19\u002FOctober-2026\u002F3.pdf",{"tldr":433,"method":434,"finding":435,"direction":144,"opportunity":436},"构建38关键点全身YOLOv8姿态框架，实现奶牛跛行自动评估并与Sprecher评分比对。","YOLOv8l-pose，12000张449头荷斯坦牛图像，120视频盲评验证。","模型mAP@0.5达83.11%，AI评分与共识Sprecher评分显著相关，AUC达0.875。","可扩展至多牧场外部验证，并结合时序行为与多模态数据提升跛行早期预警泛化性。","2026-10-05T23:30:33.217120Z",{"id":439,"title":440,"url":441,"summary":442,"summary_zh":443,"content":8,"source_name":444,"source_url":441,"published_at":445,"category":11,"cover_url":8,"hotness":12,"is_selected":52,"score":446,"score_detail":447,"sources":451,"tags":453,"primary_topic":456,"secondary_topics":459,"classification":462,"search_phrases":498,"slug":501,"view_count":186,"doi":502,"paper":503,"created_at":521},2017,"Two Decades of Computer Vision in Precision Livestock Farming: an Umbrella Review of Applications, Benchmarking Evidence, and the Performance-to-Deployment Gap Across Cattle, Pig, and Poultry Systems","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.compag.2026.112419","Computer vision is becoming a core technology in PLF, enabling non-contact monitoring, phenotyping, and decision support at both animal and group levels. This umbrella review synthesised review-level evidence from 112 reviews on cattle, pig, and poultry systems, covering visual applications for identification, phenotyping, health, behaviour, locomotion, body condition, growth, reproduction, mortality, and resource use. To preserve the review as the bibliographic unit while capturing multi-topic evidence, multi-label coding generated 328 review-domain assignments across six PLF computer vision domains. Health\u002Fstress and posture\u002Factivity were the most frequently reviewed areas; however, these patterns reflect review coverage rather than evidence strength. Benchmarking reporting averaged 11.09 out of 14 criteria, corresponding to 79.2%, but validation-critical items, including class balance, annotation protocols, validation splits, and farm or site reporting, remained less complete. Methodological confidence was limited: AMSTAR 2 classified 95 reviews as critically low and 17 as moderate, while ROBIS classified 78 as high, 17 as unclear, and 17 as low risk of bias. The corpus-level candidate-reference overlap estimate was 0.122% and is retained only as a sensitivity estimate, not as formal primary-study CCA or evidence of independence. Review-derived performance and readiness tiers identified a mean descriptive performance-to-deployment gap of 0.63 across 328 assignments. Overall, reported model performance continues to exceed documented farm-deployment evidence, highlighting the need for stronger external validation, transparent benchmarking, workflow integration, economic assessment, and user-readiness evidence before routine deployment claims can be supported.","计算机视觉正成为精准畜牧养殖（PLF）中的核心技术，能够在个体和群体层面实现非接触式监测、表型分析和决策支持。本伞状综述综合了112篇综述的综述级证据，涵盖牛、猪和家禽养殖系统，涉及识别、表型分析、健康、行为、运动、体况、生长、繁殖、死亡率和资源利用等视觉应用。为在保留综述作为文献计量单元的同时捕捉多主题证据，采用多标签编码在六个PLF计算机视觉领域中生成了328个综述-领域分配。健康\u002F应激和姿态\u002F活动是综述覆盖最多的领域；然而，这些模式反映的是综述覆盖情况而非证据强度。基准报告平均得分为14项标准中的11.09项，对应79.2%，但验证关键条目，包括类别平衡、标注方案、验证集划分以及农场或场地报告，仍不够完整。方法学可信度有限：AMSTAR 2将95篇综述评为极低质量，17篇评为中等质量；ROBIS将78篇评为高偏倚风险，17篇评为不明确，17篇评为低偏倚风险。语料库层面的候选参考文献重叠估计值为0.122%，仅作为敏感性估计保留，而非正式的一手研究CCA或独立性证据。基于综述推导的性能与就绪度分级在328个分配中识别出平均描述性性能-部署差距为0.63。总体而言，所报告的性能持续超过已记录的农场部署证据，凸显了在支持常规部署声明之前，需要更强的外部验证、透明的基准测试、工作流程整合、经济评估和用户就绪度证据。","Computers and Electronics in Agriculture","2026-09-09T00:00:00Z",87,{"impact":88,"substance":448,"depth":449,"authority":89,"freshness":90,"relevant":21,"comment":450},23,19,"基于112篇综述的伞式综述，系统揭示计算机视觉在牛猪禽精准养殖中性能与落地之间的差距，方法学与证据分级扎实，对智慧畜牧研发与部署具有较高参考价值。",[452],{"name":444,"url":441},[95,26,243,454,455],"精准畜牧","智能养殖",{"code":192,"slug":457,"name":195,"path":458,"parent_code":36},"poultry","\u002Ftopic\u002Flivestock\u002Fpoultry",[460,461],{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},{"status":45,"locked":13,"version":46,"method":47,"evidence":463},{"candidates":464,"agricultural_evidence":52},[465,470,474,479,483,489,494],{"code":39,"score":449,"supported":52,"title":466,"abstract":467,"tags":469},[405],[243,405,468],"decision support",[26,243],{"code":192,"score":89,"supported":52,"title":471,"abstract":472,"tags":473},[457],[195,457],[],{"code":36,"score":179,"supported":52,"title":475,"abstract":476,"tags":478},[315],[477],"畜牧",[],{"code":32,"score":179,"supported":52,"title":480,"abstract":481,"tags":482},[304],[304],[],{"code":484,"score":485,"supported":52,"title":486,"abstract":487,"tags":488},"C6",4,[],[454],[454],{"code":324,"score":186,"supported":52,"title":490,"abstract":491,"tags":493},[],[492],"文献计量",[],{"code":115,"score":21,"supported":13,"title":495,"abstract":496,"tags":497},[],[],[95],[499,500],"农业人工智能 计算机视觉 智慧农业 智能养殖","农业人工智能 计算机视觉","农业人工智能计算机视觉智慧农业智能养殖-2017","10.1016\u002Fj.compag.2026.112419",{"doi":502,"openalex_id":504,"authors":505,"venue":444,"cited_by_count":65,"oa_url":441,"card":516,"direction":72,"ingested_from":146},"W7212062011",[506,508,511,513],{"name":507,"orcid":8},"Alexey Ruchay",{"name":509,"orcid":510},"Vladimir Kolpakov","https:\u002F\u002Forcid.org\u002F0000-0001-9658-7034",{"name":512,"orcid":8},"Hao Guo",{"name":514,"orcid":515},"Andrea Pezzuolo","https:\u002F\u002Forcid.org\u002F0000-0002-9955-9896",{"tldr":517,"method":518,"finding":519,"direction":72,"opportunity":520},"伞式综述112篇综述，梳理牛猪禽精准养殖计算机视觉应用与部署差距。","伞式综述，多标签编码328项，AMSTAR 2与ROBIS评估方法学质量。","模型性能普遍高于农场部署证据，平均性能-部署差距0.63，验证与基准报告不足。","可研究外部验证、标准化基准与农场工作流集成，弥合性能到部署的鸿沟。","2026-09-10T23:30:01.497855Z"]