Pomegranate Leaf Disease Detection Using Supervised and Unsupervised Algorithm Techniques

Abstract

Pomegranate is a highly profitable fruit which produces abundantly in several Asian countries. However, due to a wide range of factors, the plants are infected with a wide range of illnesses that kill off the entire crop, leading to a catastrophically low product production. To address the issues in Phytopathology, this research suggests an image processing, supervised and unsupervised approach to wellness detection and categorization of pomegranate leaf. The diseases spread by insects and the environment affect the leaves. Some examples of these ailments include blight microbe, rot, and spot. The proposed system employs some images for training, and others for testing functions. First, a color image is preprocessed and employed k-means clustering to isolate the damaged region of the leaf. Locating the Region of Interest, scaling, color converting, and filtering are all part of the pre-processing process. Next, supervised learning models like ResNet and MobileNet are used to extract features and classify data. Classification criteria are utilized to locate the top-performing model. Metrics analysis reveals that MobileNet outperforms ResNet by 1.79%, providing an accuracy of 97.32%.