Non-invasive swimming speed estimation method based on tail-beat frequency determined from fish length measurement using stereo-cameras

Fish swimming speed is a standard measurement used for assessing the environment and behavior of fish in aquaculture. However, methods involving propeller-type data loggers, ultrasonic telemetry, and cameras have some limitations. Estimates of swimming speed by cameras are based on swimming trajectories; however, obtaining the swimming trajectories is difficult because the relative positions of the camera and fish body change owing to wave effects. The aim of this study was to describe and present a novel method to estimate the swimming speed of masu salmon (Oncorhynchus masou) influenced by artificially generated waves. Tail-beat frequency was estimated from periodic changes in fish length measured using a stereo camera. The swimming speed was estimated from tail-beat frequency according to the relationship between them as obtained by camera measurement; we found that a fixed stereo camera can help estimate tail-beat frequencies. The estimated data from the moving stereo camera were within the 95% prediction interval of the swimming speed estimation model constructed using the fixed stereo camera. Our study proves the potential of this method in aquaculture. Overall, this approach is a novel method to estimate the swimming speed influenced by wave conditions. In the future, this method can be automated for the identification of abnormal behaviors and detection of malformed fish.