Google has developed a new artificial intelligence tool to help marine biologists better understand coral reef ecosystems and their health, which may aid conversational efforts. The tool, SurfPerchcreated with Google Research and DeepMind, was trained on thousands of hours of reef recordings that allow reef scientists to “hear reef health from within,” track reef activity at night, and monitor reefs that are in deep or murky water.
The project began by inviting the public to listen to reef sounds over the internet. In the past year, visitors to Google By calling our website Corals they listened to over 400 hours of reef sound from locations around the world and were told to click when they heard a fish sound. This resulted in a ‘bioacoustic’ dataset focused on reef health. By crowdsourcing this activity, Google was able to create a library of new fish sounds that were used to refine its AI tool, SurfPerch. Now, SurfPerch can be quickly trained to detect any new reef sound.
“This allows us to analyze new datasets with much greater efficiency than previously possible, removing the need for training on expensive GPU processors and opening up new opportunities for understanding reef communities and their conservation,” he notes. a Google blog post about the project. The post was written by Steve Simpson, Professor of Marine Biology at the University of Bristol in the UK, and Ben Williams, a marine biologist at University College London, who study coral ecosystems with a focus on areas such as climate change and restoration.
In addition, the researchers realized that they were able to enhance the performance of the SurfPerch model by leveraging bird records. Although bird sounds and reef recordings are very different, there were common patterns between bird songs and fish sounds that the model was able to learn from, they found.
After combining Calling Our Corals data with SurfPerch in the original testsresearchers were able to reveal differences between protected and unprotected reefs in the Philippines, track restoration outcomes in Indonesia, and better understand fish community relationships on the Great Barrier Reef.
The project continues today as new audio is added to the By calling the Our Corals websitewhich will help further train the AI model, Google says.
