Assessing appetite of the swimming fish based on spontaneous collective behaviors in a recirculating aquaculture system

In order to realize the real-time appetite-based feeding in aquaculture, a novel and practical method, based on the quantification of the spontaneous collective behaviors, was proposed in this study to assess the real-time appetite of the swimming fish in a recirculating aquaculture system. First, foreground feature points of fish school were extracted using an improved complex network. Then, covariance, a modified social force model and a kinetic energy model were used to analyze the collective behaviors of the school from perspectives of dispersion degree, interaction force and the changing magnitude of the water flow field, respectively. Finally, the quantified behavioral characteristics were integrated and used to assess the appetite of fish school. The presented method shows its good performance in the expression of the collective behaviors representing five typical appetites (0.01, 0.52, 1.28, 2.26 and 2.92), and the assessing accuracy of the appetite of the school is also maintained at a low non-match rate ((2.19 ± 0.81)% best) in the context of ten different sampling durations.