智能识别技术在羽毛羽绒种类鉴别的应用研究  

Research on intelligent recognition for identification of feather and down species

作　　者：郭文 黄中浩 龙俊超 余溢镇 GUO Wen;HUANG Zhonghao;LONG Junchao;YU Yizhen(Guangzhou Inspection Testing and Certificertion Group Co,Ltd.,Guangzhou,Guangdong 511447,China)

机构地区：[1]广州检验检测认证集团有限公司,广东广州511447

出　　处：《中国纤检》2024年第8期56-59,共4页China Fiber Inspection

基　　金：广州检验检测认证集团内部科技项目“基于AI自动识别的羽绒羽毛种类分析研究”(2021kj08GF)。

摘　　要：针对现有人工辨别种类的方法存在的弊端,分析研究人工智能识别技术在羽毛羽绒种类鉴别中的应用价值,建立了人工智能识别模型,并对模型进行了训练、迭代优化,最终形成了可用于羽毛羽绒种类鉴别的识别模型。通过模型训练、模型测试和模型与人工识别时间的探究表明,模型识别的准确率大于98%,经对模型进行实际测试,其结果与人工识别法测定结果基本一致,且模型识别的时间较人工识别缩短68.8%了。Analyze and study the application value of artificial intelligence recognition technology in the identification of feather and down species,in response to the drawbacks of existing manual identification methods.Based on the characteristics of identifying down and feather species,an artificial intelligence recognition model was established,and the model was trained and iteratively optimized,ultimately forming a recognition model that can be used for identifying down and feather species.Through model training,model testing,and exploration of model and manual recognition time,the results show that the accuracy of model recognition in model training testing is greater than 98%.After actual testing of the model,the results obtained by manual recognition method are basically consistent,and the time for model recognition is shortened by 68.8%compared to manual recognition.

关 键 词：人工智能 识别 鸭 鹅 种类鉴别 测定 

分 类 号：TS107[轻工技术与工程—纺织工程]