Visual-Based System for Fish Detection and Velocity Estimation in Marine Aquaculture

As global aquaculture continues to expand to meet the rising seafood demand, optimization of feeding remains a crucial issue for the industry to address to achieve sustainable development.This study proposed a visual-based system for estimating fish velocity, which is to be integrated into a farmer's feeding operation to determine the optimal feed amount.The YOLOv8 algorithm was utilized to detect fish in underwater videos, enabling precise monitoring of fish behavior.The results indicate a successful fish detection with an accuracy of 85%.The fish velocity estimation approach demonstrated the difference between the hungry fish and the normal fish behavior.The findings suggested that integrating fish velocity data into feeding operations can significantly enhance feed efficiency, reduce waste, and promote sustainable aquaculture practices, ensuring optimal fish growth while minimizing environmental and economic impacts.