You might roll your eyes as you watch the drone take off into the sky and hover over the Australian coast, the camera pointed straight down at the sparkling turquoise waters. “Another TikTok influencer trying to get the perfect shot,” you mutter to yourself. But if you look closely at the pilot, you’ll see that there’s a sign next to him that reads “Keep Clear” in bright yellow and red letters. This is not a TikTok influencer.
It’s an Australian surf lifesaver using the above drone to spot sharks on the beach before they get too close to swimmers like you. The New South Wales government has agreed to invest millions of dollars (more than A$85 million, to be precise) in shark control measures over the next few years in order to achieve better coexistence with these predators. Sure, there’s helicopter patrols and the ever-splitting shark nets and drum lines, but a 2020 survey showed that drone-based shark surveys were what the public preferred when it came to their safety.
This is not news as the state government has been using drones to detect sharks since 2016 and has partnered with Surf Life Saving NSW to continue the work since 2018. The drones climb to a height of over 60 meters. Piloted by stranded surf lifeguard pilots to maneuver the technology across the blue ocean while watching a live video feed of what the drone is seeing. They’re not just gazing at the beautiful waves – they’re chasing the shape of sharks that swim beneath the surface. While pilots are trained in how to distinguish shark shapes from one another (a great white shark looks different from a tiger shark, for example) and between other animals (a shark looks different from a seal or a large fish), it can be a bit difficult , when the weather conditions are not the best. The wind could be making it difficult to get a clear picture, the sunlight hitting the wrong spot, or the water is just too murky and dark and there are tons of algae everywhere…
Drone pilots generally make the right decision 60% of the time – which is both reassuring and a bit unnerving for some. That’s why a team of scientists decided to find out if artificial intelligence (AI) is the answer. dr Cormac Purcell received a grant from the NSW Department of Primary Industries to carry out this research at Macquarie University with Dr. Paul Butcher, a fellow at Southern Cross University and Deakin University. Together, the team set out to build the “most robust” AI detector for sharks – and test it right off the waters of Australia. You see, while most AIs do reasonably well in the lab, they face a number of challenges in the real world, which is why Purcell and Butcher were keen to test their detector in the wild. “Initial results from previous AI-powered shark detection systems suggest 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 difference on NSW beaches has been a challenge. […] In essence, Machine Learning Operations explicitly recognizes that AI-driven software requires regular updates to maintain its effectiveness.”
Researchers were able to create a surfing lifesaver mobile app by meticulously tracking and identifying sharks to feed the information to the AI software so it could “learn,” as the authors explain: “With this new dataset, we trained.” machine learning model to detect ten species of marine life, including various species of dangerous sharks such as great white shark and whaler. And then we embedded that model into a new mobile app that can highlight sharks in live drone footage and predict the species. We have worked closely with the NSW Government and Surf Lifesaving NSW to trial this app on five beaches over the summer of 2020.”
How did it go? “Our AI Shark Detector did pretty well. Dangerous sharks were identified frame by frame under realistic conditions 80% of the time. We made a conscious effort to make our tests harder by challenging the AI to work with invisible data taken at different times of the year or from different looking beaches.” Now there were some limitations for the app – for example For example, sharks with similar outlines were difficult to identify, while smaller animals were difficult to spot. However, the team is confident that the AI ”is now mature enough to be used in drone-based shark-spotting operations on Australian beaches.” But unlike regular software, it needs to be monitored and updated frequently to maintain its high reliability in detecting dangerous sharks.”
With summer down under, the drones are out again, scanning the waters and hopefully keeping beachgoers safe. “AI can play a key role in making these flights more effective, enabling greater reliability in drone surveillance, and eventually leading to fully automated shark-sighting operations and reliable automated alerts,” the authors conclude.