A SCALABLE IMAGE-BASED FRAMEWORK FOR DETECTING AND MONITORING RICE LEAF DISEASES

Abstract

Rice is one of the world's most important staple crops, feeding more than half of the global population. However, rice production is significantly affected by leaf diseases such as bacterial leaf blight, blast, brown spot, and tungro, resulting in substantial yield losses. Traditional disease diagnosis relies on manual field inspection, which is time-consuming, subjective, and unsuitable for large-scale monitoring. This paper proposes a scalable image-based framework for automatic rice leaf disease detection and monitoring using deep learning and cloud-enabled analytics. The proposed framework integrates image preprocessing, data augmentation, lightweight convolutional neural networks (CNNs), transfer learning, and IoT-enabled monitoring to provide accurate disease identification in realtime. A MobileNetV3-EfficientNet hybrid architecture optimized using Bayesian hyperparameter tuning is employed to classify healthy and diseased rice leaves. Extensive experiments demonstrate that the proposed framework achieves an overall accuracy of 98.94%, precision of 98.71%, recall of 98.66%, F1-score of 98.68%, and AUC of 99.31%, outperforming existing deep learning models. The proposed scalable architecture enables deployment on smartphones, drones, and edge devices for smart agriculture applications.