A groundbreaking project to explore the potential of training artificial intelligence (AI) models for detecting sea urchins is about to start, with the goal of transforming how Southeast Australia’s shallow rocky reef ecosystems are monitored and managed.
The project is led by researchers at QUT, in partnership with Wilderness Wildcatch — a Mallacoota-based seafood company committed to marine stewardship. Supported by funding from the Fisheries Research and Development Corporation (FRDC), the initiative targets the long-spined sea urchin, Centrostephanus rodgersii.
Although Centrostephanus rodgersii is native to Australian waters, its population has expanded significantly along the east coast, raising serious concerns for reef health and resilience.
The species has been highlighted in the 2023 Senate inquiry into invasive marine species due to its ecosystem-altering impacts. By overgrazing seaweeds and other marine species, Centrostephanus rodgersii creates “urchin barrens” — stretches of seafloor reduced to bare rock — which dramatically reduce reef biodiversity and productivity. These shifts also pose a direct threat to high-value commercial fisheries such as abalone.
“There’s a long-standing need for reliable, scalable tools to monitor the abundance and movement of Centrostephanus,” commercial diver Reiner Hurst, who is representing Wilderness Wildcatch, said.
“While divers provide valuable observational data, there are limits due to depth, time underwater, and positional accuracy year-on-year.”
The first stage of the project will involve local divers capturing underwater imagery to help train AI models capable of automatically detecting and counting sea urchins. If successful, these models could be integrated into autonomous underwater vehicles (AUVs) or remotely operated vehicles (ROVs), paving the way for large-scale, high-resolution monitoring of reef ecosystems.
“We’re building on our strong foundation of AI research applied to marine ecosystems,” said Professor Matthew Dunbabin, senior researcher and robotics specialist at QUT.
“Our previous work using AI for delivering baby corals to degraded reefs and detecting and tracking Crown-of-Thorns Starfish on the Great Barrier Reef has shown how powerful these technologies can be in supporting marine management. By applying similar tools to the challenge of monitoring Centrostephanus, we aim to generate high-resolution, unbiased data that will help managers and communities better understand reef health and where targeted interventions are most needed.”
