Improving plant disease classification with deep learning based prediction model using explainable artificial intelligence detection

Plant diseases remain a major challenge for sustainable agriculture, often causing significant yield losses and economic setbacks. This work presents an advanced AI-driven system for plant disease classification and detection using state-of-the-art convolutional neural networks (CNNs) and object detection models. For classification, hybrid models such as NASNetMobile and Xception achieved outstanding performance, with NASNetMobile attaining 99.5% accuracy and Xception scoring 98.9%, outperforming traditional architectures like ResNet50 and MobileNet. For object detection, YOLO-based models were evaluated, with YOLOv5s6 demonstrating the highest mean Average Precision (mAP) of 78% and superior recall for real-time applications. These results confirm the system's robustness and scalability for accurately identifying plant diseases in diverse environments. The proposed approach empowers farmers with a practical, automated solution for early detection and intervention, contributing to improved crop health and food security.