LCNFN: LeNet-Cascade Neuro-Fuzzy Network for Grape Leaf Disease Segmentation and Multi-Classification

Abstract

Grapes are the most productive fruit; however, they are at risk for a greater number of diseases. Grapes are one of the finest popular fruits and prime things for wine production; consequently, their relent and grape quality are of extensive monetary value. Nevertheless, the grape leaves are vulnerable to several diseases, which are affected by weather conditions and their atmosphere, and they are majorly affected by fungi, viruses and bacteria. Moreover, diverse conventional approaches have neglected to classify grape leaf disease. To subdue this gap, an effectual module is presented for the multi-classification of grape leaf disease utilising the LeNet-Cascade Neuro-Fuzzy Network (LCNFN). The original image of the grape leaf is filtered by applying a Laplacian filter and region of interest (ROI) extraction. The black spot segmentation is performed by Black Hole Entropic Fuzzy Clustering (BHEFC), and then feature extraction is progressed. Thus, multi-classification is performed with LCNFN, which is classified into Isariopsis leaf spot, black spot, black measles and healthy. The measures used for LCNFN are accuracy, sensitivity and specified observed 89.6%, 91% and 91.2%, respectively.