Ocean Vision AI is leveraging artificial intelligence and machine learning to advance our ability to process underwater imagery. This initiative paves the way for creating reproducible data pipelines for running and validating models while engaging with non-experts to enhance ocean literacy. Ocean Vision AI has recently developed an open-source database for annotating ocean imagery on a platform called FathomNet. In addition, the project is in the process of designing a video game for ocean enthusiasts to interact with ocean imagery and improve machine learning models, as well as a portal framework for practitioners to work with and label objects in imagery. Both of these tools contribute to creating and curating training data that is vital for automated detection and classification of objects in imagery. This work is funded by the National Science Foundation (NSF) as part of its Convergence Accelerator program.
This effort has brought together expertise from CeNCOOS, Climate Change AI, CVision AI, Ocean Discovery League, the University of California, Santa Cruz and more.
Stay up to date on this project on Twitter: @OceanVisionAI.