You might be rolling your eyes as you see the drone take off to the skies and hover over the Australian coastline, camera angled straight down towards the glistening turquoise water. “Another TikTok influencer trying to get the perfect shot,” you grumble to yourself. But if you look closely at the pilot, you’ll notice they’ve got a sign next to them that says “Keep Clear” in bright yellow and red letters. This is no TikTok influencer.
It’s an Australian surf lifesaver, using the above drone to spot sharks at the beach before they get too close to swimmers like yourself. The New South Wales government has positioned itself to invest millions of dollars (more than A$85 million to be exact) in shark mitigation measures over the next couple of years in an effort to better co-exist with these predators. Sure, there are helicopter patrols and the ever-divisive shark nets and drumlines, but a 2020 survey showed drone-based shark surveys was what the public preferred when it came to their safety.
This isn’t news, as the state government has been using drones to spot sharks since 2016, and teamed up with Surf Life Saving NSW to continue the work since 2018. The drones soar to a height of over 196 feet (60 meters), piloted by rrained surf lifesaving pilots to maneuvre the technology over the blue ocean while watching a live video feed of what the drone sees. They’re not just looking at the gorgeous waves – they’re on the hunt for the shape of sharks swimming under the surface. While the pilots have training on how to tell a shark shapes between one another (a great white shark looks different than a tiger shark, for example) and between other animals (a shark looks different than a seal or big fish), it can be kind of tricky when weather conditions aren’t the best. The wind could be making it hard to get a clear picture, the sun glare is hitting just the wrong spot, or the water is just too murky and dark and there’s tons of seaweed everywhere…
Drone pilots generally make the right call 60% of the time – which is both reasurring and a bit nail-biting for some. Which is why a team of scientists decided to see if artificial intelligence (AI) was the answer. Dr. Cormac Purcell received funding from the NSW Department of Primary Industries to conduct this research at Macquarie University, alongside Dr. Paul Butcher, an adjunct of Southern Cross University and Deakin University. Together, the team set out to build the ‘most robust’ shark AI detector possible – and test it right off the waters of Australia. See, while most AI does reasonably okay in the lab, they have a bunch of challenges to deal with in the real world, which is why Purcell and Butcher were keen on testing their detector in the wild. “Early results from previous AI-enhanced shark-spotting systems have suggested the problem has been solved, as these systems report detection accuracies of over 90%,” the authors write in The Conversation. “But scaling these systems to make a real-world difference across NSW beaches has been challenging. […] Essentially, machine learning operations explicitly recognises that AI-driven software requires regular updates to maintain its effectiveness.”
The researchers were able to create a mobile app for surf lifesavers by painstakingly tracking and identifying sharks to feed the information to the AI software so it could ‘learn,’ as explained by the authors: “Using this new dataset, we trained a machine learning model to recognise ten types of marine life, including different species of dangerous sharks such as great white and whaler sharks. And then we embedded this model into a new mobile app that can highlight sharks in live drone footage and predict the species. We worked closely with the NSW government and Surf Lifesaving NSW to trial this app on five beaches during summer 2020.”
How did it fare? “Our AI shark detector did quite well. It identified dangerous sharks on a frame-by-frame basis 80% of the time, in realistic conditions. We deliberately went out of our way to make our tests difficult by challenging the AI to run on unseen data taken at different times of year, or from different-looking beaches.” Now there were some limitations to the app – for example, for sharks that had similar outlines, it was hard to ID them, while smaller animals were hard to detect. But the team is confident that AI is “now mature enough to be deployed in drone-based shark-spotting operations across Australian beaches. But, unlike regular software, it will need to be monitored and updated frequently to maintain its high reliability of detecting dangerous sharks.”
With it being summer down under, the drones are out to scan the waters once again and, hopefully, protect beachgoers. “AI can play a key role in making these flights more effective, enabling greater reliability in drone surveillance, and may eventually lead to fully-automated shark-spotting operations and trusted automatic alerts,” the authors conclude.