Eyes in the Sky: Using Drones and AI to Understand Beach Use on the Gold Coast

Australia’s beaches are thriving with activity. From surfers on the ocean to joggers on the sand, our beaches host millions of visitors each year, making them an important recreational and economic asset. Estimating the number of people that use these natural spaces, and understanding how they use them, is essential to help plan public safety measures and the maintenance of infrastructure and amenities. 

However, assessing the number of beach visitors can be challenging. Traditional methods at ground level, such as static cameras and headcounts performed by lifeguards, are limited by their precision and the area they can cover, leading to significant underestimates. Instead, researchers have taken to the skies, using drones and artificial intelligence (AI) to accurately monitor beach use, helping to improve coastal management and public safety. 

From Drones to Data

Drone surveys are a promising approach for monitoring beach use as they provide a more comprehensive and accurate picture of the coastline. However, the large amount of video imagery that needs to be processed and analysed is a significant challenge.

Fortunately, advances in AI and computer vision have allowed for the automated identification, counting and monitoring of objects in videos and photos with extremely high accuracy. 

Researchers from Griffith University, in collaboration with the City of Gold Coast, assessed – for the first time – beach use through a combination of drones and AI. The study conducted over 500 drone surveys across 29 beaches and 37 km of coastline. The researchers investigated various environmental conditions, times, days of the week, and seasons to capture variations in beach use. 

Professor Rod Connolly, Director of Griffith University’s Coastal and Marine Research Centre and a co-author in the study, said, “Important community decisions can only be made if there’s useful data. Without knowing what people are doing and in what number, councils can’t plan for the future. That’s really what it boils down to: planning infrastructure so that it’s cost effective, and making sure that people are safe.”  

The study showed that drones are a cost-effective and robust way to monitor beach use, covering half a kilometre of beach in just a few minutes. The research team used 2 AI algorithms: one to distinguish between people on land and in water, and another to categorise activities of beachgoers, such as walking, resting, or swimming. The algorithms achieved approximately 90% accuracy in their classifications.

The AI algorithms used in the study were modified from FishID – a tool used to identify, count and measure aquatic animals and plants, developed with co-investment from the ARDC. By using the same software protocols and tweaking the equations, Professor Connolly and his team were able to repurpose the FishID tool to count beachgoers rather than aquatic life. The AI algorithms were run on GPUs provided by the ARDC Nectar Research Cloud. 

“We’ve been using Nectar for two things. One is the modelling and AI, which is important because it requires a lot of computing power. The second is to store the significant amount of video footage, which has been extremely helpful for all our projects,” said Professor Connolly.

Insights into Beach Use on the Gold Coast

The study estimated that there were about 34 million visitors to the Gold Coast beaches between 2022 and 2023, about twice the number estimated by traditional lifeguard headcounts. “A lifeguard’s main job is not to count people, but to save lives,” noted Professor Connolly when explaining the difference between the two counts.  

The study also revealed interesting insights into how beachgoers used the beach, with 55% of visitors on land and 45% in the water. Unsurprisingly, the highest visitor counts were observed in ‘hotspot’ beaches such as Burleigh Heads, Surfers Paradise and Coolangatta. 

“One of the big concerns was whether there are spots where people are swimming that are not near a life-saving tower,” said Professor Connolly. “That was the information that the managers really homed in on. They looked at each bump in the frequency plot and asked, ‘Do we need to investigate? Do we need to reconsider the position of towers?’”

Shaping Safer Beaches and Smarter Coastal Management

Findings from the study prompted the Gold Coast City Council to increase patrols along busy beaches to improve public safety. The new approach also helps to better allocate funds for beach protection, assets, and amenities by providing more accurate beach use data. 

After 3 years of beach data collection, applications of the drone and AI technology are now being expanded. 

“Behind the beaches are hundreds of kilometres of waterways, canals, rivers and lagoons – some artificial and some natural. The City of Gold Coast has asked us if we can apply the drone and AI technology to understand how these are all being used,” said Professor Connolly.  

By combining the power of drones, AI, and the Nectar Research Cloud, Australia’s beaches, and the people who cherish them, stand to benefit from data-driven decision-making for years to come.  

Read the journal paper by Herrera et al.: Drone Insights: Unveiling Beach Usage through AI-Powered People Counting. 

Learn how to use the ARDC’s National GPU Service.

FishID received co-investment (doi.org/10.47486/PL071) from the ARDC. The ARDC is enabled by the National Collaborative Research Infrastructure Strategy (NCRIS).

Written by Dr Cintya Dharmayanti, Scientell. Reviewed by Prof Rod Connolly (Griffith University) and Jo Savill, Dr Paul Coddington and Kerry Levett (ARDC).