The system uses high-quality cage camera images and artificial intelligence to monitor fish behaviour and allow real-time adjustment of the feeding rate.
Victor Valerio, general manager of Puerto Varas-based 3se, said: “After three years of intensive work that merged the experience of the team in production, artificial vision, artificial intelligence, models developed in other realities involving human behaviour, among other things, the Intelligent Feeding system is ready to be applied in real management of feeding in the salmon industry, minimising feed losses, facilitating operator work and controlling the efficiency and efficacy of the process.”
Installed in Chile
In fact, said Valerio, “we are already have the system installed and functioning in Chilean companies, where they have been able to evidence its benefits”.
As well as monitoring feeding, the fish behaviour registered is useful for analytics that continually improve the system. User-friendly dashboards support feed operators with indicators and alarms to prevent insufficient or excess feeding, says 3se.
The company said the initiative has put emphasis on fish behaviour, developing a complex and multiple system of algorithms that can interpret several individual and collective fish reactions that allow for optimised feeding, and provide valuable additional information for fish health and general fish condition.
Feeding stopped ‘before pellets are visible’
Intelligent Feeding is based on the use of high-resolution cameras, usually already present in the cages, to detect individual and collective fish behaviour, along with other environmental parameters of interest detected by sensors in cages.
Algorithms are used to analyse the data to determine if fish are hungry or satiated, to avoid both over-feeding and under-feeding.
Valerio said the system could “stop feeding even before pellets are visible on the screens indicating satiety of fish”.
Andre Flem, a senior executive and entrepreneur in the salmon industry and a partner and director of 3se, said: “This is a very reliable and efficient system with multiple projections, because it learns and improves based in AI, as well as capturing and processing information to detect in advance risky diseases and abnormal signs that deserve immediate attention. This is why we are also promoting the system in other countries, particularly Norway.”