A machine learning and citizen science approach to analyse underwater footage from Sweden’s first marine national park.
Underwater footage displaying the outputs of a machine learning model that identifies cold-water corals.
Jannes Germishuys/ Combine
With over 6,000 different marine seaweeds and animals, Kosterhavets National Park is the most species-rich marine environment in Sweden.
Remotely Operated Vehicles
For the last 30 years, researchers have used Remotely Operated Vehicles to monitor the park.
The footage helps biologists understand threats like climate change and the positive effects of protecting the national park.
Fuzzy images and the high animal diversity makes the manual classification of underwater videos a challenging task for researchers.
Wildlife.ai is developing an open-source approach to analysing large amounts of subsea movie data for marine ecological research.
The approach incorporates three distinct modules to: manage and archive the subsea movies, involve citizen scientists to accurately classify the footage and train and test machine learning algorithms for detection of biological objects.
Koster Seafloor Observatory 2.0
The Koster Seafloor Observatory 2.0 has new movies of wonderful seafloor critters, extensive identification keys, and teaching modules for students. Underwater video of the artificial intelligence algorithm [...]
Koster Observatory Initial Results
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 [...]