Spatial behavioral characteristics and statistics-based kinetic energy modeling in special behaviors detection of a shoal of fish in a recirculating aquaculture system

In global aquaculture, fish welfare is attached increasing importance to. Fish behavior holds important information of its living welfare. Thus, in order to detect special behaviors concerned with fish welfare in a recirculating aquaculture system (RAS), a modified kinetic energy model of the whole shoal instead tracking individuals inside a shoal was proposed in this study. The dispersion, velocity and turning angle of the shoal were regarded as the key factors in this model. First, with the reflective regions of the water surface taken into account, two alternative ways were presented to extract the necessary spatial behavioral characteristics of the shoal. Then, optical flow, entropy and statistics were adopted to measure the dispersion status (dispersion of individuals in space) and the changing magnitude of the motion (behavioral characteristics) of the shoal. Finally, the modified kinetic energy model was obtained by combining the dispersion measurement with the changing magnitude of the behavioral characteristics. Referring to the human observation and the evacuation of the gastro-intestine contents-based long experiment of Nile tilapia (Oreochromis niloticus) in RAS, the proposed model shows good performance in detection of the emergent gathering and scattering behaviors with 97.20 ± 1.23% success rate at least and 0.61 ± 0.08% missing report rate at most.