Sharks and Drones: How SharkEye Uses Drones to Keep People Safe from Great Whites
BY Zacc Dukowitz26 November 2020
Drone videos like the one below, which shows sharks swimming dangerously close to beachgoers, have become fairly common over the last few years.
These videos are enthralling—not to mention disturbing—but new research is taking the potential of drone technology for capturing sharks beyond the realm of entertainment, using it as a tool to spot great whites and other dangerous sharks to help keep people safe.
A project called SharkEye is one of the leaders in this effort.
SharkEye is based at the Benioff Ocean Initiative at the University of California, Santa Barbara. Researchers involved in the project are using drones to monitor the ocean for great whites, and developing a system that combines Artificial Intelligence (AI) and drone technology to create automated reports about the presence of sharks in the water.
The project has two goals: to learn more about shark biology and to help keep people safe.
About the SharkEye Program
Researchers in the SharkEye program feed raw drone footage of the ocean into a computer model they have trained with machine learning to recognize great white sharks.
They are working on ways to combine the visual data collected by drone with relevant information on the temperature of the ocean and migratory patterns in order to create predictions for the likelihood of great whites showing up on a certain day, near a certain section of the beach.
Image credit: SharkEye
The SharkEye project is a collaboration of SalesForce AI research, computer scientists at San Diego State University, and marine biologist from U.C. Santa Barbara.
It’s an interesting time in the United States. There’s now more great white sharks spotted off the coast than we’ve ever seen before and there’s not a great resource for understanding those populations.
– Michael Jones, Product Management—Senior Director at SalesForce
The SalesForce component may seem a little out of left field, but Dr. Douglas McCauley, the director of the Benioff Ocean Initiative, also works with AI researchers at SalesForce (Marc Benioff, who funds McCauley’s lab, is the founder and CEO of SalesForce), so the partnership is actually pretty organic.
For now, the SharkEye system is still in development.
Actual shark-tracking efforts conducted by the SharkEye team are manual, following this routine:
- A drone pilot flies a drone along a preprogrammed path at 120 feet in the air above the ocean, filming as it goes.
- A second drone follows, taking a more meandering route, also filming the ocean.
- The pilots monitor their video feeds in real time. If a shark is spotted, they send a text to the people signed up to get shark alerts.
This manual approach to monitoring the waters has been used by SharkEye at Padaro Beach over the last two summers, a location that is both a popular surfing destination and a nursery for juvenile great whites.
Eventually, the text alerts may become formalized as automated alerts that go out on various channels—text, social media, or local news—modeled on surf reports, letting people know what the shark forecast is for the day.
The number of sharks spotted off the coast of Southern California has been increasing in recent years, creating a potentially urgent need for the kind of technology SharkEye is developing.
Learn more about the SharkEye program in this video:
Australia’s Shark-Tracking Project
SharkEye is not the only one working on ways to automate shark-tracking using drones and AI.
While SharkEye is still developing its AI system, researchers in Australia completed a trial back in June of AI software made to track sharks with drone data at five different beaches, using it successfully to identify and track the presence of several species of sharks.
Photo credit: Surf Life Saving NSW
The software was trained using machine learning, leveraging three years of visual data collected manually by drone from a previous program that tracked sharks in real time. The project was a collaboration between Surf Life Saving New South Wales (NSW) and the NSW Department of Primary Industries (DPI).
Our team has been developing a new machine-learning algorithm that runs on mobile devices to identify shark species in real time from drones.
– Dr. Cormac Purcell, Project Lead
According to the researchers leading the project, it was the first independent, scientific trial of an AI algorithm for detecting shark species by drone.
What do you think—will there be an autonomous shark report generated via drone data in our lifetime? Share your thoughts in this thread on the UAV Coach community forum.