Abstract

Vegetation health prediction is one of the most important research questions in precision agriculture and natural resource management. While multispectral and hyperspectral imagery are commonly used, their high cost limits widespread adoption. Unmanned aerial vehicles (UAVs) equipped with standard RGB cameras provide a cost-effective alternative for high-resolution vegetation monitoring. In this study, we propose a deep learning framework to predict vegetation health directly from UAV RGB images. We extract both spatial and spectral representations to model subtle variations in canopy reflectance and leaf coloration associated with plant vigor and stress conditions. Comprehensive experiments demonstrate that the proposed deep learning model achieves competitive accuracy compared to conventional vegetation index-based approaches, offering an efficient, scalable, and economical solution for automated vegetation health monitoring in agricultural and ecological applications.