Enhancing Early Plant Disease Detection with the Advanced GNN-based Model Leveraging Hyperspectral Imaging

Accurate early detection of plant diseases remains a critical challenge due to the complexity of hyperspectral data and the limitations of traditional machine learning models. This study introduces a novel Graph Neural Network (GNN)-based framework that effectively models spectral-spatial correlations, enabling superior disease identification from hyperspectral images. Unlike conventional Convolutional Neural Networks (CNNs) and classical machine learning approaches, the proposed model leverages advanced graph representations and attention mechanisms to capture intricate spectral dependencies at both local and global scales. Evaluated on a diverse dataset of over 5,000 labeled hyperspectral images covering multiple plant diseases and species, our model achieves 92% accuracy, 90.5% precision, 91% recall, and a 90.7% F1-score, significantly outperforming CNN-based methods and classical classifiers like Support Vector Machines (SVMs) and Random Forests. The integration of hyperspectral feature extraction with GNN-based architectures enhances the model’s ability to discern subtle spectral variations associated with early disease onset. This study establishes a new benchmark in hyperspectral plant disease detection, demonstrating the transformative role of graph-based deep learning in precision agriculture. The proposed framework offers a scalable, efficient, and accurate solution, paving the way for next-generation AI-driven agricultural monitoring systems aimed at improving global food security.