A Review of Research on Fish Feeding Behavior Based on Computer Vision

With the development of aquaculture, monitoring fish feeding behavior is crucial for breeding efficiency and resource utilization. Traditional manual observation methods are inefficient and subjective, while computer vision technology provides a new solution for real-time quantitative analysis of fish feeding behavior. This article summarizes the key steps in monitoring fish feeding behavior, including fish species selection, image information collection methods, and details various quantitative methods applicable to feeding behavior analysis, such as area method, behavior feature statistics, image texture features, and Delaunay triangulation method. These methods each have their own advantages and can adapt to different research needs. Deeply understanding and quantifying the feeding patterns of fish not only helps optimize aquaculture management and improve resource utilization, but also provides scientific basis for ecological protection.