NDVI-Based Machine Learning System for Precision Agriculture

Abstract

AgriSmart is a machine learning-based platform designed to assist farmers and agricultural planners in monitoring crop health, predicting crop yield, and recommending optimal fertiliser usage without relying on IoT sensors. The system leverages satellite and drone imagery to analyse crop conditions using vegetation indices, such as NDVI, in combination with historical crop, soil, and weather data. Machine learning models, including regression and convolutional neural networks, are employed to forecast crop yield, classify crop health, and provide actionable fertiliser recommendations. An interactive dashboard visualises crop health heatmaps, predicted yields, and fertiliser recommendations, enabling data-driven decision-making. This approach is sustainable, scalable, and fully implementable on a personal computer, promoting efficient resource use and early intervention to prevent crop stress and maximise productivity.