A vision transformer-based deep learning approach for lemon leaf disease detection

Abstract

Early detection and effective management of lemon leaf diseases play a critical role in modern agricultural practices. This study explores the potential of the Vision Transformer (ViT) model for classifying lemon leaf diseases, evaluating the success of deep learning-based approaches in this domain. A comprehensive performance analysis was conducted to assess the model's ability to accurately distinguish between various disease types. Experimental findings demonstrate that the ViT model outperforms other models with an accuracy rate of 99.32%. Furthermore, the results surpass previously reported accuracy rates in the literature by 0.76%, proving the proposed method to be more effective than existing approaches. The model's performance was evaluated in detail based on classification accuracy, confirming that the ViT model offers high precision in detecting lemon leaf diseases. The findings of this study highlight the critical importance of automation in agricultural disease detection and contribute significantly to disease management processes by reducing the need for manual observation. Early detection of diseases enables more targeted interventions, reduces unnecessary chemical usage, promotes environmental sustainability, and enhances crop productivity.