Coral reefs are some of the most diverse ecosystems on the planet. However, Australia's Great Barrier Reef has seen dramatic coral losses in the last 30 years, with other reefs experiencing similar declines. Climate change, pollution, and overfishing are major contributors. In Australia, outbreaks of the coral-eating crown-of-thorns starfish have been shown to cause significant coral loss.

Healthy reefs are critical to fisheries and food security. They also provide countless additional benefits by protecting coastlines from storm surges, elevating tourism-based economies, and pushing forward drug discovery research.

Controlling invasive species outbreaks is critical to promote coral growth and resilience, so Google teamed up with Australia's Commonwealth Scientific and Industrial Research Organisation (CSIRO) to tackle this problem with advanced technology.

Machine learning can scale and improve underwater surveys, monitoring and mapping out these harmful invertebrates in real time to help control teams address and prioritize outbreaks. Google used internal compute resources to experiment with different modeling and training techniques that improved accuracy and performance.

Google is working with CSIRO to deploy the model using local hardware, leveraging a graphics processing unit (GPU) connected to a GoPro to map out the harmful starfish in real time. Our goal is to empower environmental scientists and researchers to effectively detect and prioritize outbreaks. We will publish our model and accompanying training to help organizations around the world use this technology to protect their coral reefs.