Multi-Model Citrus Leaf Disease Classification and Feature Interpretation Using Deep Learning and Explainable AI

Abstract In this paper, we propose an advanced GUI-based system for citrus plant healthclassification by comparing deep learning and traditional machine learning models using leafimages. The system integrates a Convolutional Neural Network (CNN), Support Vector Machine(SVM), Random Forest (RF), K-Nearest Neighbors (KNN), and Logistic Regression (LR) todetect diseases such as Black Spot, Melanose, Canker, Greening, and Healthy leaves. Theinterface supports image input, prediction, model comparison, and SHAP-based featureimportance visualization. The proposed system achieved highest prediction accuracy of 88%using Random Forest, while CNN achieved 62%. Additionally, feature importance visualizationsrevealed insights into model interpretability. This framework offers a valuable tool for precisionagriculture, supporting disease diagnosis and model transparency.