Hybrid Approach of Cotton Disease Detection for Enhanced Crop Health and Yield

Abstract

The well-being of cotton crops is of utmost importance for maintaining agricultural productivity, and the early detection of diseases plays a critical role in achieving this objective. This study introduces a comprehensive approach for creating a machine learning-based system capable of identifying diseases in cotton plants through the analysis of leaf images. The research encompasses stages such as acquiring the dataset, pre-processing the data, training the model, developing an ensemble model, evaluating the models, and analyzing the results. Several machine-learning models are trained and evaluated to determine how well they can classify cotton leaves as "Healthy" or "Diseased." These models include Random Forest, Support Vector Machine (SVM), Multi-Class SVM, and an Ensemble model. This investigation yields a practical and visually informative system for disease detection, which can contribute to disease prevention, thereby enhancing both crop yield and quality. This work underscores the significance of continuous improvement by periodically updating the models and explores the potential of advanced techniques such as deep learning.