AI-driven hake catch monitoring: pioneering size-based inventory control in longline fishing

Abstract

This study introduces a transformative approach in longline fisheries, employing YOLO v7 object detection algorithm for real-time, automated sizing of hake. We have developed an artificial intelligence (AI) model based on Yolo v7 that classifies captured specimen of hake into four commercial size categories, applicable to recorded or live-streaming video from on-board cameras in a new electronic monitoring (EM) system concept called iObserver Lite. This dual applicability demonstrates the model’s adaptability to different operational scenarios. Obtained results reveal the YOLO v7-based model’s outstanding accuracy in hake size detection, maintaining high precision in both controlled and real daily fishing activity environments. This performance is pivotal for real-time inventory management and offers the potential for advanced fishery analytics and real-time fish auction sales even before landing the catches. Moreover, by providing instantaneous catch size data, the technology aids in optimizing fishing efforts and supports sustainable fishing practices. The integration of YOLO v7 in longline fishing represents a significant technological leap, enhancing operational efficiency and contributing to achieving sustainable fishery management soon. This breakthrough showcases the vast potential of artificial vision and AI in revolutionizing the fishing industry, heralding a new era towards efficiency and sustainability of fishing activity regarding marine resource exploitation.