How can we tell if cultivated fish are stressed when their living conditions change?  GuangXv Wang, a researcher at project partner China Agricultural University, is helping answer this question by investigating the impacts of low temperature in recirculating aquaculture systems on juvenile fish behavior using state-of-the-art neural network and deep learning technology.

Fish are exposed to low temperatures and the behavior response is recorded in short video clips. These clips are then split into frames and give information about the trajectory and interactions between individual fish in a tank that can be detected and modeled by complex algorithms. This ultimately allows AquaDetector researchers to determine the degree and onset of stress behavior, and provides valuable insight on fish welfare.