Health Classification of Rice Plants Based on UAV Remote Sensing Using Random Forest Algorithm

Abstract

Bogor Regency acts as a central hub for rice production in West Java, yet frequent disease outbreaks often jeopardize the consistency of agricultural yields. Farmers struggle with these plant diseases because the infections frequently result in significant crop losses or total harvest failure. The immense size of paddy fields makes manual monitoring methods inefficient, driving a requirement for automated systems to monitor crop health across large areas. The current research focuses on building a classification model that identifies whether rice plants are healthy or diseased using aerial photographs. The process utilizes drone-based remote sensing technology where the data is analyzed using the Random Forest algorithm. Final model evaluations show solid performance with an accuracy of 85% and a precision of 100%. The system also achieved a recall of 70% and an F1-Score of 0.82. Evidence suggests that the Random Forest algorithm works effectively to separate healthy rice from diseased crops using drone imagery. Farmers can use such technological approaches as practical tools to detect diseases early and manage their fields better.