An Intelligent Framework for Grassy Shoot Disease Severity Detection and Classification in Sugarcane Crop

Abstract

The Grassy Shoot Disease is a severe problem in sugarcane crops, affecting their productivity and causing significant economic losses. The research aims to introduce a model that utilizes both CNN and SVM techniques to make precise predictions about the severity levels of Grassy Shoot Disease in sugarcane cultivation. The methodology involves data preprocessing, CNN-based feature extraction, SVM-based classification, and model evaluation. The data preprocessing phase involved data cleaning, normalization, and augmentation, followed by the extraction of features using a three-layer CNN model. Following feature extraction, the extracted features were fed into an SVM-based classifier with regularisation to avoid overfitting. The classifier's overall accuracy was 81.53%, and its precision, recall, F1-score, and support values ranged from 65.71% to 85.37% depending on the severity level. These results show that the suggested method is a solid method for accurately estimating the degrees of Grassy Shoot Disease severity in sugarcane crops.