Smart Aquaculture: Integrating Computer Vision and IoT for Automated Crayfish Health Assessment via Federated Learning

The rising global demand for aquaculture products necessitates sustainable alternatives to traditional farming methods that often rely on manual monitoring prone to errors and inefficiencies. This research presents a novel federated learning framework integrated with computer vision and IoT technologies for automated crayfish health assessment in precision aquaculture. The system employs edge devices equipped with IR, NoIR, and standard cameras to acquire images from 1.67m² fiber tank cultivation systems, processed using YOLOv11 for real-time crayfish detection, density estimation, and disease identification. A key innovation is the implementation of federated learning, which enables collaborative model training across multiple farms while preserving data privacy, addressing a critical barrier to technology adoption in the aquaculture industry. The framework achieved 87.4% detection accuracy for density estimation, 88.5% accuracy in growth stage classification, and demonstrated proof-of-concept disease detection capabilities. IoT sensors continuously monitor water quality parameters with data processed locally using edge devices and synchronized with AWS cloud services when connectivity permits. Field trials with Australian Blue Crayfish (Cherax quadricarinatus) at aquafarm validated the framework’s effectiveness in reducing labor costs while maintaining data privacy through federated learning that reduced bandwidth usage by 90%. The system generates actionable insights through an interactive dashboard, enabling data-driven decision-making for feeding schedules and harvest planning, offering a scalable solution for sustainable aquaculture development in resource-constrained environments.