期刊
JOURNAL OF DISCRETE MATHEMATICAL SCIENCES & CRYPTOGRAPHY
卷 23, 期 1, 页码 273-282出版社
TARU PUBLICATIONS
DOI: 10.1080/09720529.2020.1721890
关键词
Deep Learning; Multiclass SVM; Convolutional Neural Networks; Plant Diseases; Disease Severity; Tomato Late Blight; Agriculture
资金
- Department of Science and Technology (DST), Government of India, New Delhi, under Interdisciplinary Cyber Physical Systems (ICPS) Programme (Project Tilted Application of Internet of Things (IoT) in Agriculture Sector) [T-319]
For food security in future, precise measurements of disease incidence and severity are crucial for suitable treatments and adopting preventive measures. In this paper, the authors have implemented three well known CNN models, namely, AlexNet, SqueezeNet and Inception V3, for evaluating disease severity in Tomato Late Blight disease. The images utilized were selected from the PlantVillage dataset and separated into three stages (early, middle and end) of disease severity. The CNN architectures were implemented in two different modes, i.e. transfer learning and feature extraction (where the extracted feature set was used to train a multiclass SVM). As compared to the other two networks, AlexNet achieved the highest accuracy in both approaches, 89.69% and 93.4% respectively.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据