Explainable and Fair Leaf Disease Detection using Multimodal Deep Learning

Abstract

Agriculture plays a crucial role in global food security, and early detection of plant leaf diseases is essential to prevent crop losses and improve yield quality. Traditional disease detection methods rely heavily on manual inspection, which is time-consuming and error-prone. This paper proposes X-FairLeaf, an explainable and fairness-aware multimodal framework for plant leaf disease detection. The system integrates image, environmental, temporal, and relational data to improve detection accuracy. Explainable AI techniques such as SHAP and attention maps provide interpretability, while fairness constraints ensure unbiased predictions across plant species. Experimental results show improved accuracy and robustness compared to baseline models.