Citizen scientists have analysed +50,000 videos and trained a machine learning model to recognise cold water corals!

Underwater video of the artificial intelligence algorithm recognising a cold water coral in real-time.
Victor Anton /

Peer-reviewed article

We are happy to announce the publication of the full description of our open-source citizen science and machine learning system to recognise marine species.

Automatic identification of corals

One of the first results of the Koster seafloor observatory is the ability to recognise key ecological species from seafloor imagery, like the cold water coral Lopehlia pertusa.

The presence of cold water corals is one of the reasons why the National Park was established. This species of coral has been in decline for more than a decade and we still haven’t understood the causes behind it.

Thanks to thousands of citizen scientist annotations, we have trained an artificial intelligence algorithm that can recognise this species of coral.

The algorithm is allowing us to automatically process thousands of videos collected over the past 20 years so that we can better understand the cause (or causes) for the coral’s decline.

Photo of cold water coral (Desmophyllum pertusum)

Desmophyllum pertusum.
Tomas Lundälv and Lisbeth Jonsson / Kosterhavest National Park

Analysis app

The machine learning model is available in our free app where you can upload your own footage.

You can manipulate the classification parameters and see how it influences the model accuracy.

Share this story!