CNN —
On summer mornings, local kids like to gather at California’s Padaro Beach to learn to surf the gentle rapids. A few years ago, the beach even became a popular spot for young great white sharks.
This prompted the Benioff Institute for Marine Sciences (BOSL) at the University of California, Santa Barbara to launch SharkEye, an initiative that uses drones to monitor what’s happening beneath the ocean’s surface.
When a shark is sighted, SharkEye sends a text message to about 80 people who have signed up for the alerts, including local lifeguards, surf shop owners and parents of children taking lessons.
In recent years, authorities and lifeguards from New York to Sydney have been using drones to keep swimmers safe and monitor camera feeds. Pilots must battle choppy waters and glare from the sun, and stay focused on their screens to distinguish between sharks and paddleboarders, seals, and rippling seaweed. One study found that human-supervised drones only detect about 60 percent of sharks.
Both a research program and a community safety tool, SharkEye uses the videos it collects to analyze shark behavior and feeds the footage into a computer vision machine learning model — a type of artificial intelligence (AI) technology that allows computers to glean information from images and videos — to train it to detect great whites near Padaro Beach, near Santa Barbara.
“Having automated shark detection could be really helpful, not just here in California, but for a lot of communities,” Neil Nathan, project scientist at BOSL, who earned a master’s in environmental studies from Stanford University a few years ago, told CNN.
The growing popularity of drones and the proliferation of social media may make it seem like sharks are everywhere, but warmer oceans are pushing them into new habitats, exacerbating the situation as young whites, which grow to about 8 to 10 feet in length, prefer to stay closer to shore, making them more visible to beachgoers.
But shark attacks are rare: 69 people worldwide were bitten by unprovoked sharks in 2023, roughly the same as the annual average of 63 from 2018 to 2022. Only 10 of those were fatal, according to the Florida Museum of Natural History’s International Shark Attack File.
Although no fatal attacks have been recorded at Padaro Beach, some local residents became concerned when sharks began wandering there.
So for about five years, SharkEye has been sending drones out regularly to monitor the coastline, once spotting 15 juvenile great whites in a single day.
Early tests have shown the AI technology is already working “amazingly well,” spotting most sharks that humans can, as well as some that humans miss (perhaps because they swim deeper and are less easily spotted), Nathan said.
This summer, the project began field testing its technology by pitting drone pilots against an AI: The pilots survey an area and count the number of sharks they find, and SharkEye’s models then analyze the video to see how many sharks they can find.
Currently, community alerts are based on human analysis. If all goes well, Nathan said, by the end of the season or early next summer, these reports could become AI-assisted (with manual monitoring and checking). In the future, the process could become fully automated, making it faster and more accurate.
AI and Wildlife
AI technology is being leveraged in a variety of ways to mitigate human-wildlife conflict: in India, AI-enabled cameras are alerting villagers when a tiger is approaching livestock, and in Australia, the technology is being used to control dangerous animals.
Ripper and academics developed what is said to be the first shark-identification algorithm used on a drone several years ago, and the latest version of the software is being tested in Queensland, Australia, Mexico and the Caribbean to detect sharks and crocodiles.
But AI isn’t yet widely used to detect sharks. Surf Life Saving New South Wales, which protects dozens of beaches along the state’s coast, including Sydney’s iconic Bondi Beach, uses drones at 50 locations, but a spokesperson told CNN that their drones don’t currently use AI.
A group at an Australian university working on developing an AI-powered shark-spotting tool wrote in 2022 that the technology could struggle when it encounters conditions that aren’t present in the training data.
SharkEye plans to make its models freely available so researchers can modify and improve them, and develop an AI-powered app that makes it easy for lifeguards, drone enthusiasts and others to process the footage — not only keeping people safer, but helping humans better understand and protect sharks.
Nathan said it’s too early to say how much retraining will be required to expand Shark Eye to other areas, but he hopes there won’t be as many issues in other parts of California with similar coastlines, as long as drone pilots fly at the same speeds and altitudes.
Honolulu officials said this month they were considering starting a drone shark monitoring program, local media reported. If SharkEye’s technology is used in places like Hawaii, where tiger sharks are most of concern, further retraining may be required because the water is a different color. But Nathan said SharkEye is open to working with other regions to help adapt the model.
“We want to arm our communities with the knowledge and awareness to feel safer sharing the waters with these creatures,” Nathan says. “Sharks are amazing creatures and we’re always learning new things.”
