Solving underwater fish segmentation with realistic labeled data generation and test-time augmentation

In this video, Andrei Jelea from NORCE shows how AI can solve one of aquaculture’s big challenges: segmenting fish in underwater videos. By generating realistic synthetic training data and using test-time augmentation, the method reduces the need for manual labeling and improves monitoring of fish health and behavior.

Kategorier

• Havbruk til havs
• Teknologi