Adaptive artificial multiple intelligence fusion system (A-AMIFS) for enhanced disease detection in Nile Tilapia

Abstract

The development of an efficient disease detection system for Nile Tilapia is critical due to the significant economic and food security impacts of disease outbreaks. Traditional disease identification methods are labor-intensive, inefficient, and often fail to detect early signs of disease, leading to substantial economic losses. This study introduces the Adaptive Artificial Multiple Intelligence Fusion System (A-AMIFS), an advanced model that innovatively combines image augmentation, ensemble image segmentation methods, and ensemble Convolutional Neural Network (CNN) architectures. The system utilizes a non-population-based artificial multiple intelligence system (np-AMIS) for optimizing image augmentation and a population-based system (Pop-AMIS) for decision fusion, demonstrating superior performance. Evaluated on two novel datasets, Nile Tilapia Disease-1 (NTD-1) and Nile Tilapia Disease-2 (NTD-2), the system achieved an overall accuracy of 98.26 %, precision of 98.35 %, recall of 98.30 %, and an F1-score of 98.32 %, significantly outperforming existing methodologies. Additionally, a "chatbot" feature was developed to enable farmers to automatically detect fish diseases using the ensemble model as the backend classification system, achieving an impressive classification accuracy of over 98 %. These results underscore the system's robustness in detecting various diseases in Nile Tilapia and its potential to transform disease detection in aquaculture. The proposed system reduces manual labor, optimizes disease identification processes, and enhances disease management strategies, promoting more sustainable and productive aquaculture practices. This research highlights the indispensable role of AI techniques in overcoming the complex challenges of disease detection and management in aquaculture, presenting efficient and effective disease management practices. By leveraging advanced image augmentation, ensemble segmentation methods, and ensemble CNN architectures, this study presents a revolutionary approach to disease detection in Nile Tilapia. The integration of a user-friendly chatbot interface further enhances the technology's accessibility and practical application, empowering farmers to proactively manage disease outbreaks and mitigate economic losses.