Hard attention mechanism integrated with VGG16 for high-precision tomato leaf disease classification

Abstract Tomato diseases represent a major threat to agricultural productivity and global food security. Early and accurate identification of these diseases is essential for effective crop management and yield preservation. In this study, we propose a hard attention–enhanced VGG16 model for automatic tomato leaf disease identification. By integrating a hard attention mechanism into the VGG16 architecture, the proposed model is able to focus on the most informative regions of the input images, thereby improving its capability to discriminate between healthy and diseased tomato leaves. The proposed approach is evaluated using a publicly available dataset of tomato leaf images. Experimental results demonstrate that the hard attention–based VGG16 model significantly improves classification performance compared to the standard VGG16 architecture and several existing deep learning models. In particular, the proposed model achieves an accuracy of 96.0%, outperforming the baseline VGG16 model, which achieves 92.3% accuracy on the same dataset. Furthermore, the integration of the attention mechanism enhances the interpretability of the model by highlighting the image regions that contribute most to the prediction process. This explainability provides valuable insights into the model’s decision-making process, making it a reliable and transparent tool for intelligent plant disease identification systems.