Play Live Radio
Next Up:
Available On Air Stations

Central Coast researchers use artificial intelligence for ocean exploration

Screen Shot FathomNet.png
Katija, K., Orenstein, E., Schlining, B. et al.
Sci Rep 12, 15914 (2022)
FathomNet: A global image database for enabling artificial intelligence in the ocean.

The vast amount of underwater data now available from autonomous vehicles and ocean sensors is overwhelming for scientists sorting through it, so a team from Monterey Bay Aquarium Research Institute (MBARI) is using artificial intelligence (AI) to develop algorithms that will make the process easier. Algorithms are the guidelines used to program computers and make them smart.

“I want intelligent vehicles to go out into the ocean to scan and look for new life,” MBARI engineer Kakani Katija said.

Katija and her team received a grant from the National Science Foundation to develop a new database called FathomNet for collecting and labeling underwater imagery.

She said there is a lot we don’t know about the ocean and its inhabitants, and AI is a powerful tool for underwater exploration.

“Scientists estimate that anywhere from 30 to 60% of life in the ocean has yet to be discovered,” Katija said. 

She uses existing images to build algorithms for the ocean robots. She said the more images she collects, the better the algorithms.

“I’m throwing [in] lots of different images of a jellyfish or lots of different images and views of a shark, and together the algorithm can start to pick up features that differentiate those animals – it can say, with some relative certainty, it is a jellyfish, or it is a shark,” she said. 

Katija said the technology also allows scientists to monitor the impacts of climate change or other threats to the ocean.

Joost Daniels © 2019 MBARI
MBARI Principal Engineer Kakani Katija inspects the Mesobot, a new generation of underwater robot, during first field trails on board MBARI’s research vessel Rachel Carson in Monterey Bay. Mesobot is being developed by engineers at Woods Hole Oceanographic Institute and MBARI.

“Robotic vehicles can be monitoring a location for a long time. We can have a persistent observational presence at a place to let us know when something is different or has changed,” Katija explained.

The FathomNet database is open to scientists, ocean industries, and the general public. The project team is also developing a video game.

“Through video gaming, we can invite people to participate in this exploration of the ocean. You could be the first person to look at this footage of an animal that was completely unknown to science,” she said.

Pokémon Go is an example of the type of interactive game Katija said she wants to create for ocean exploration.

“A game like Pokémon Go completely changed behaviors of people as they were looking for fake animals, and how can we change behavior through games or gaming around animals that live here on our planet that we don’t know very much about,” she said.

The video game is scheduled for release next year. She said the goal for players is to add to the knowledge bank while also becoming good stewards of the ocean.

You can find out more about FathomNet on

Beth Thornton is a freelance reporter for KCBX, and a contributor to Issues & Ideas. She was a 2021 Data Fellow with the USC Annenberg Center for Health Journalism.