Australian scientists built an AI that analyses coral reefs about 700 times faster than people doing it by hand. It.s open access, and it.s now spreading across the Pacific and Southeast Asia.
What it is
ReefCloud was built by the Australian Institute of Marine Science (AIMS) in Townsville, with support from the Australian Government. It’s an open-access platform that analyses underwater photos of reefs from anywhere.
How it works
Divers and scientists upload photos from reef surveys. Machine learning reads each image, classifies what’s in it by things like colour, shape and size, and standardises the data so you can track how a reef is changing over time. The models are trained on local conditions, such as water clarity and the species found in each region.
The numbers
- 80 to 90% accuracy analysing reef composition
- About 700x faster than the traditional method, where scientists go through each image manually
Where it’s used
ReefCloud has been tested and rolled out with partners including Palau, Fiji and the Maldives. Through a Southeast Asia project it’s being used in Brunei, the Philippines and Vietnam, and the Philippines has adopted it as its standard monitoring method.
My read
Built in Townsville, used across the region. This is what AI was supposed to be for: taking slow, manual science work and making it fast enough to keep up with what’s happening to the reefs.