<![CDATA[Background. Crop disease is a major threat to smallholder farmers who lack access to timely diagnostic tools. Traditional detection methods rely on visual inspection and often occur too late to prevent significant yield losses. Early detection using hyperspectral imaging and artificial intelligence presents a transformative solution for precision agriculture in resource-limited settings. Purpose. This study aims to develop and evaluate an AI-assisted early detection system for crop diseases using hyperspectral imaging and deep learning, tailored for application in smallholder farms. Method. A convolutional neural network (CNN) model was trained on hyperspectral data collected from five farm sites, with ground-truth annotations by agricultural experts. The model’s performance was evaluated using standard classification metrics, including accuracy, precision, recall, and F1-score. A case study was also conducted to assess real-world applicability. Results. The model achieved an average detection accuracy of 94.2% across all locations, with F1-score reaching 0.92 when using hyperspectral features. Confusion matrix analysis indicated high true positive and true negative rates, confirming reliability. In a field case, early diagnosis enabled targeted intervention and improved yield by 22% compared to prior seasons. Conclusion. The integration of hyperspectral imaging and deep learning offers a practical and scalable solution for early disease detection in smallholder farms. The system demonstrates high accuracy, adaptability, and operational feasibility in real-world conditions. Future work should focus on expanding crop and disease types, user interface development, and integration with mobile and IoT-based platforms.]]>
Copyrights © 2025
