{"title":"Trialling一种用于识别新西兰小蠊科的卷积神经网络","authors":"D. Ward, Brent Martin","doi":"10.3897/jhr.95.95964","DOIUrl":null,"url":null,"abstract":"Computer vision approaches, such as deep learning, potentially offer a range of benefits to entomology, particularly for the image-based identification of taxa. An experiment was conducted to gauge the ability of a convolution neural network (CNN) to identify genera of Braconidae from images of forewings. A deep learning CNN was trained via transfer learning from a small set of 488 images for 57 genera. Three-fold cross-validation achieved an accuracy of 96.7%, thus demonstrating that identification to genus using forewings is highly predictive. Further work is needed to increase both the coverage to species level and the number of images available.","PeriodicalId":50185,"journal":{"name":"Journal of Hymenoptera Research","volume":null,"pages":null},"PeriodicalIF":1.4000,"publicationDate":"2023-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Trialling a convolution neural network for the identification of Braconidae in New Zealand\",\"authors\":\"D. Ward, Brent Martin\",\"doi\":\"10.3897/jhr.95.95964\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Computer vision approaches, such as deep learning, potentially offer a range of benefits to entomology, particularly for the image-based identification of taxa. An experiment was conducted to gauge the ability of a convolution neural network (CNN) to identify genera of Braconidae from images of forewings. A deep learning CNN was trained via transfer learning from a small set of 488 images for 57 genera. Three-fold cross-validation achieved an accuracy of 96.7%, thus demonstrating that identification to genus using forewings is highly predictive. Further work is needed to increase both the coverage to species level and the number of images available.\",\"PeriodicalId\":50185,\"journal\":{\"name\":\"Journal of Hymenoptera Research\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":1.4000,\"publicationDate\":\"2023-02-17\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Hymenoptera Research\",\"FirstCategoryId\":\"97\",\"ListUrlMain\":\"https://doi.org/10.3897/jhr.95.95964\",\"RegionNum\":3,\"RegionCategory\":\"农林科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"ENTOMOLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Hymenoptera Research","FirstCategoryId":"97","ListUrlMain":"https://doi.org/10.3897/jhr.95.95964","RegionNum":3,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENTOMOLOGY","Score":null,"Total":0}
Trialling a convolution neural network for the identification of Braconidae in New Zealand
Computer vision approaches, such as deep learning, potentially offer a range of benefits to entomology, particularly for the image-based identification of taxa. An experiment was conducted to gauge the ability of a convolution neural network (CNN) to identify genera of Braconidae from images of forewings. A deep learning CNN was trained via transfer learning from a small set of 488 images for 57 genera. Three-fold cross-validation achieved an accuracy of 96.7%, thus demonstrating that identification to genus using forewings is highly predictive. Further work is needed to increase both the coverage to species level and the number of images available.
期刊介绍:
The Journal of Hymenoptera Research is a peer-reviewed, open-access, rapid online journal launched to accelerate research on all aspects of Hymenoptera, including biology, behavior, ecology, systematics, taxonomy, genetics, and morphology.
All published papers can be freely copied, downloaded, printed and distributed at no charge for the reader. Authors are thus encouraged to post the pdf files of published papers on their homepages or elsewhere to expedite distribution. There is no charge for color.