Abstract
<jats:p>Potato is one of the most important food and cash crops in Bangladesh, contributing significantly to food security and the national economy. However, potato production is severely affected by various leaf diseases, which reduce crop yield and quality when not detected at an early stage. Traditional disease diagnosis mainly depends on manual observation by agricultural experts, making the process time-consuming, expensive, and less accessible for farmers in remote areas. This study proposes an explainable deep learning-based framework for multi-class potato leaf disease detection using real-world field images collected from agricultural environments in Bangladesh. The framework employs deep learning models to classify potato leaf images into multiple disease categories while integrating Explainable Artificial Intelligence (XAI) techniques to improve the transparency and interpretability of model predictions. By highlighting disease-affected regions, the proposed system enables users to understand the reasoning behind each prediction and supports informed decision-making. The developed approach aims to enhance diagnosis accuracy, reduce crop losses, minimize unnecessary pesticide use, and promote smart farming practices. The proposed framework provides a practical, reliable, and farmer-friendly solution for intelligent agricultural disease management in Bangladesh.</jats:p>