A Hybrid System for Pigeon pea Leaf Disease Detection and Classification (PLDDC) Using CNN and SVM

Abstract

Pigeon pea is a multipurpose leguminous plant that could be used as human food and livestock feed.  It is an ideal crop for the semi-arid areas of Asia, Africa and America and there is great potential for it to be more widely grown. Plant diseases are the most significant factor affecting productivity, and they must be addressed to maintain their value. The authors aim to identify this essential factor influencing plant growth, stressing the importance of early plant disease diagnosis in order to address the underlying causes of the diseases at an early stage. As a result, in this research work implemented hybrid model that combines Convolutional Neural Networks (CNN) and Support Vector Machines (SVM) to identify three common diseases, leaf webber, leaf spots and sterilic mosaic, and classify images into four groups that include both the disease-related and healthy images. After extracting utilitarian characteristics from the input data using CNN, an optimal Support Vector Machine classifier is used to classify the features, and the accuracy of the results is recorded. The proposed framework illustrated an overall accuracy of 93.75%.