Stressed fish tend to become unhealthy fish, so it's very important that fish farm operators spot stress as early as possible. A new AI-based system could help them do so, by monitoring the rate at which each fish breathes.
The technology is being developed by scientists from Norway's SINTEF research organization and the Norwegian University of Life Sciences (NMBU). In its present prototype form, it incorporates an underwater camera which shoots horizontally into a water-filled tub containing seven salmon.
AI algorithms on a linked computer analyze the camera's output in real time, identifying each salmon by its distinguishing features, and monitoring the rate at which each fish opens and closes its mouth. That mouth action corresponds to its rate of respiration, which increases with stress levels.
In lab tests, the AI simultaneously watched over nine of the tubs as the water temperature was increased to warmer-than-ideal levels, and as the oxygen levels were decreased. In both cases, the algorithms reliably detected the salmons' increased respiratory rates that accompanied their increase in stress.
The scientists state that one of the system's big selling features is the fact that it monitors all of the fish in an enclosure, as opposed to a single "sentinel fish" which may not be a good indicator of how the others are faring.
"We found that the breathing rates of the individual fish vary," says Dr. Judit Vas, a behavioral biologist at NMBU. "Some of them breathe more slowly or faster than others. The newly developed camera technology, which monitors each salmon individually, is therefore of great help."
Two papers on the research, which is part of the Biorelevans project, were recently published in the journals Aquaculture and IFAC-PapersOnLine.
There's more information on the system in the following video.
Sources: Norwegian SciTech News, Biorelevans
