利用人工智能技术进行蠓翅图像识别  

IMAGE RECOGNITION OF MIDGE WING BASED ONARTIFICIAL INTELLIGENCE

作　　者：苏影[1] 柯乔 李婷婷 刘德星 廉国胜[1] 成子栋 SU Ying;KE Qiao;LI Ting-ting;LIU De-xing;LIAN Guo-sheng;CHENG Zi-dong(Zhuhai International Travel Healthcare Center(Port Clinic of Gongbei Customs District),Zhuhai 519020,Guangdong,China;School of Public Health,Fudan University,Shanghai 200032)

机构地区：[1]珠海国际旅行卫生保健中心(拱北海关口岸门诊部),广东珠海519020 [2]复旦大学公共卫生学院,上海200032

出　　处：《寄生虫与医学昆虫学报》2024年第4期224-229,共6页Acta Parasitologica et Medica Entomologica Sinica

摘　　要：目的 利用人工智能技术探索建立蠓类的自动分类识别方法,以实现蠓虫自动分类。方法 对珠海口岸3种蠓科昆虫尖喙库蠓、台湾蠛蠓、异域库蠓进行图像采集,然后分别以深度学习方法(建立在PaddlePaddle框架支持下的VGG卷积神经网络模型)和传统的机器学习方法(利用Matlab软件建立以手工设计特征为输入的BP神经网络模型)进行图像识别。结果 成功建立了基于VGG模型和BP模型的蠓虫自动分类技术。测试集在VGG模型的预测正确率为100.0%,在BP模型的预测正确率为94.7%,两种方法无显著性差异(P>0.05)。结论 人工智能技术应用于蠓翅图像识别可取得良好效果。Objective The aim of the study is to establish an automatic method for the recognition of common midges based on artificial intelligence technology.Methods Wing images of three species of midges from Zhuhai port,including Culicoides oxystoma,C.peregrinus,and Lasiohelea taiwanai were subjected to develop a reliable automatic recognition method based on a deep learning method.The VGG convolutional neural network model based on the PaddlePaddle framework and the BP neural network model based on Matlab software with manual design features were involved for image recognition.Results Automatic recognition based on VGG model and BP model were established with the prediction accuracy of the test set in the VGG model and BP model was 100.0%and 94.7%,respectively,with no significant difference(P>0.05).Conclusions The artificial intelligence technology can achieve reliable species recognition based on image recognition of midge wings.

关 键 词：蠓 图像分类 VGG模型 BP模型 

分 类 号：R384.5[医药卫生—医学寄生虫学]