In the past, there has been technology that counts the number of fish that pass through the salmon ladder past the power plant. With the help of AI for image recognition, it will now be possible to register many more factors that will increase knowledge and improve the biological diversity of fish life.
- As there is a lack of functioning fishing routes, there is a high probability that many fish ladders will be built in the future. We see an opportunity to use this technology to learn more about fish migration and for our fish ladders to be effective, says Mats Billstein who is responsible for research for hydropower at Vattenfall.
A system that teaches itself
Vattenfall is behind the project, which involves training a system that reads the fish with the help of cameras to recognize certain factors. This is done by telling the system what should be recognized by the fishermen and doing it enough times for the system to learn itself.
In the first step that is being tested now, it is a matter of registering whether the salmon is wild-born or farmed. When the fat fin on the farmed fish is cut off, the system is taught to read whether it is present or not. The system will also be able to read the size and whether it carries fungal disease.
The big challenge
It is planned for further steps together with Luleå University of Technology and SLU, the Swedish University of Agricultural Sciences, which will make it possible to also register species, sex and what is described as the great challenge, to recognize the fishermen as unique individuals.
The idea is that the first step will be operational during this year.
In the clip above, you get a glimpse of how the technology works and hear the researcher Mats Billstein talk about the benefits.