[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-2612":3},{"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,"view_count":32,"doi":8,"paper":33,"created_at":42},2612,"AgroVisNet: A lightweight Convolutional Network and the BD-PlantDX Expert-Validated Benchmark for Radish, Potato and Pointed Gourd Disease Classification","https:\u002F\u002Fzhichai.net\u002Ftopic\u002F178634732","Mandal等发表在arXiv:2609.10469。提出AgroVisNet一种从头训练的紧凑卷积网络和BD-PlantDX专家验证基准(12,432张田间图像，涵盖孟加拉国Bogura和Nilphamari地区收集的萝卜、马铃薯和尖瓜12类健康和病害状态)。AgroVisNet仅290,572个可训练参数在BD-PlantDX上达到99.52%测试准确率和加权F1，超过所有六个ImageNet预训练轻量级主干网络，同时参数量少8.7到16.8倍。",null,"arXiv 2609.10469","2026-09-09T01:00:00Z","论文",10,false,73,{"impact":16,"substance":17,"depth":18,"authority":19,"freshness":20,"relevant":21,"comment":22},15,22,18,12,6,1,"提出仅29万参数的轻量卷积网络与专家验证的12类作物病害基准，准确率99.52%且显著优于预训练主干，方法新颖、数据规模扎实，对农业AI病害识别有实用参考价值。",[24],{"name":9,"url":6},[26,27,28,29,30,31],"智慧农业","农业人工智能","马铃薯","萝卜","病害识别","轻量卷积网络",0,{"doi":8,"openalex_id":8,"authors":34,"venue":8,"cited_by_count":32,"oa_url":8,"card":35,"direction":39,"ingested_from":41},[],{"tldr":36,"method":37,"finding":38,"direction":39,"opportunity":40},"提出轻量卷积网络AgroVisNet及专家验证的萝卜、马铃薯和尖瓜病害图像基准BD-PlantDX。","从头训练紧凑CNN，使用12,432张田间图像、12类病害，与6个预训练轻量主干","仅29万参数即达99.52%准确率，超越所有ImageNet预训练轻量模型且参数少8.7-16.8倍","农业人工智能与决策模型","可探索跨地区跨作物泛化、田间复杂背景鲁棒性及模型轻量化部署到移动端的研究。","agent","2026-09-16T00:03:52.515327Z"]