Classification of Shallot Leaf Health Based on RGB Images Using Machine Learning Algorithms

Abstract

Agricultural productivity often declines due to plant diseases that reduce yield and quality. Shallot (Allium ascalonicum L.) is one of Indonesia’s key horticultural commodities, yet it is highly vulnerable to leaf diseases such as purple blotch and Fusarium-induced moler. This study aims to develop a shallot leaf health classification model using RGB images and machine learning algorithms. The proposed system employs an experimental approach based on a publicly available image dataset consisting of three categories: healthy, purple blotch, and moler-infected leaves. Preprocessing stages include image resizing, noise reduction, and contrast enhancement to improve visual clarity. Feature extraction combines RGB color histograms and Gray-Level Co-occurrence Matrix (GLCM) texture descriptors to obtain informative features. Two algorithms Support Vector Machine (SVM) and Random Forest (RF) were trained and evaluated using accuracy, precision, recall, and F1-score metrics. The results show that both models achieved perfect classification performance, with RF demonstrating slightly higher stability and robustness. These findings confirm that the integration of RGB imagery and lightweight machine learning algorithms provides a reliable, low-cost, and computationally efficient solution for early detection of shallot leaf diseases. The proposed approach contributes to precision agriculture development and offers potential deployment for smallholder farmers through simple, camera-based monitoring systems.