Feeding behavior quantification and recognition for intelligent fish farming application: A review

Experience-driven manual and mechanical feeding often fail to accurately address fish appetite needs in aquaculture. Analyzing the relationship between fish behavior and feeding activity through efficient modeling to develop intelligent feeding systems is a key challenge in current research. In this regard, computer vision has emerged as an effective tool for behavior analysis due to its cost-effectiveness and non-invasive nature. Based on this, this paper provides a comprehensive review of the research progress in applying computer vision to fish feeding behavior analysis and its practical implementations. Firstly, this study introduces the key indicators reflecting fish feeding states. Subsequently, we summarize and discuss the critical methods developed in recent years for quantifying and recognizing fish feeding behavior using computer vision. Additionally, the paper consolidates and analyzes case studies of intelligent feeding applications based on these methods. Finally, based on current research advancements, this paper highlights four key challenges future directions for intelligent feeding technology, including real-time performance, stability, efficiency, and fish welfare. The goal of this work is to offer a scientific foundation and valuable reference for the development of intelligent feeding systems in aquaculture, while also inspiring further reflection and exploration among researchers.