Considerable variation in Atlantic salmon morphology creates challenges for developing reliable and agreed criteria for manual identification of farmed escapees. It is therefore desirable to develop a digital tool that can provide a precise and objective classification of salmonids with associated uncertainty . Thus, the IdentiFish project was established to develop a general and precise machine vision to distinguish between escaped and wild salmon.
Based on a unique data set collected through a n ational m onitoring p rogram for escaped farmed salmon in Norwegian water courses and a commercial salmon fishing app Elveguiden , the project group will develop a model for distinguishing escaped farmed salmon from wild salmon using machine vision . While t he national monitoring program surveys ~200 rivers annually, Elveguiden collaborates with over 500 management teams and landowners and has over 55,000 registered anglers on the platform.
The machine vision model will be made publicly available on the Institute of Marine Research’s website, and at the same time implemented in Elveguiden’s app. An app capable of distinguishing between escaped and wild salmon using machine vision can contribute to more precise monitoring of escape events. Immediate classification via app provides opportunities for rapid reporting to management authorities which can reduce the reaction time to implement necessary measures, such as targeted recapture as well as limit the extent of the escape event and thus minimize financial losses.
The project is an example of how artificial intelligence can be used to strengthen monitoring of wild salmonids and contribute towards sustainab le aquaculture production. The project is funded by the Norwegian Seafood Research Found (FHF 901937) and is a collaboration between researchers, developers, and entrepreneurs: Norwegian Institute of Marine Research, Elveguiden (https://elveguiden.no/en) , Norwegian Veterinary I nstitute and Norwegian Institute for Nature Research.