Sanjida Sultana Reya, Md Abdul Malek, Anik Debnath
{"title":"Deep Learning Approaches for Cabbage Disease Classification","authors":"Sanjida Sultana Reya, Md Abdul Malek, Anik Debnath","doi":"10.1109/ICRPSET57982.2022.10188553","DOIUrl":null,"url":null,"abstract":"Cabbage diseases such as black rot, downy mildew, and white rust are frequent and have a negative impact on yield. However, existing research lacks an accurate and rapid detector of cabbage diseases to assure healthy cabbage production. In this research, the transfer learning approach has been employed for many state-of-the-art CNN architectures, such as VGG16, VGG19, mobilnetv2, and InceptionV3, to determine the most optimal solution for this problem. A dataset of around 1500 images from three different classes is employed to train and validate the models. Among the multiple CNN models evaluated, vgg16 produced 95.55% test accuracy, which is far superior to other similar experiments conducted recently.","PeriodicalId":405673,"journal":{"name":"2022 International Conference on Recent Progresses in Science, Engineering and Technology (ICRPSET)","volume":"83 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-12-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 International Conference on Recent Progresses in Science, Engineering and Technology (ICRPSET)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICRPSET57982.2022.10188553","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Cabbage diseases such as black rot, downy mildew, and white rust are frequent and have a negative impact on yield. However, existing research lacks an accurate and rapid detector of cabbage diseases to assure healthy cabbage production. In this research, the transfer learning approach has been employed for many state-of-the-art CNN architectures, such as VGG16, VGG19, mobilnetv2, and InceptionV3, to determine the most optimal solution for this problem. A dataset of around 1500 images from three different classes is employed to train and validate the models. Among the multiple CNN models evaluated, vgg16 produced 95.55% test accuracy, which is far superior to other similar experiments conducted recently.