REA-CNN: A Region-Aware & Enhanced Attention CNN for Robust Leaf Disease Classification

Automated plant leaf disease detection using deep learning has achieved high accuracy on benchmark datasets; however, its performance often degrades when applied to real-world agricultural images affected by background clutter, illumination variability, and partial occlusions. These factors limit the reliability of conventional convolutional neural network (CNN)–based models trained under controlled conditions. To address this limitation, this paper proposes a Region-Aware and Enhanced Attention Convolutional Neural Network (REA-CNN) for robust classification of plant leaf diseases. The proposed framework integrates explicit region-aware pre-processing for background suppression with an attention-enhanced CNN backbone, enabling the model to focus on disease-relevant visual patterns. Unlike conventional two-stage CNN–SVM pipelines, REA-CNN is trained in a fully end-to-end manner, allowing joint optimization of feature extraction, attention refinement, and classification. Experimental results show that the proposed approach achieves higher classification accuracy and improved generalization on heterogeneous and real-world images compared to existing methods. These results demonstrate the effectiveness of combining region awareness and attention-guided learning for developing practical and deployable decision-support systems in precision agriculture.