A feature ranking-based deep learning secure framework for multi-class leaf disease detection Online publication date: Mon, 27-Jun-2022
by M. Nagageetha; N.V.K. Ramesh
International Journal of Ad Hoc and Ubiquitous Computing (IJAHUC), Vol. 40, No. 1/2/3, 2022
Abstract: Multi-class leaf disease prediction is one of the challenging tasks in large image databases due to uncertainty and high dimensional feature space. Most of the traditional deep learning frameworks are used to classify a single class disease prediction with limited feature space. However, these frameworks have high false positive rate and error rate due to various background, noisy appearance and semantic high- and low-level features for classification problem. Also, feature extraction and classification are the major problems in traditional convolution neural network (CNN) on multi-class leaf datasets. In this work, a novel image feature extraction-based deep learning classifier is designed and implemented on large multi-class leaf datasets. In this framework, a hybrid statistical leaf shape extraction method is used to find the essential features in the conditional probability-based principal component analysis (BPCA) approach. A novel deep learning classifier is proposed to improve the leaf disease prediction rate with high true positivity and accuracy on the multi-class leaf disease datasets. Experimental results show that the present framework has high computational performance than the traditional deep learning frameworks for multi-class classification.
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Ad Hoc and Ubiquitous Computing (IJAHUC):
Login with your Inderscience username and password:
Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.
If you still need assistance, please email subs@inderscience.com