INDUSTRY-LEADING TECHNOLOGY
ReefSound has 3 products for monitoring coral reef health; industry-leading AI software, remotely-operated underwater vehicle (ROV), and cost-effective ReefBox
THE REEFSOUND AI
Groundbreaking
AI technology
ReefSound’s core product is an ensemble of 2 AI models. Users input one-minute long coral recordings, which the AI identifies as either healthy or degraded.
We trained our AI with a dataset sourced from a research institute in Sesoko Island, Japan, which contains hours of sound recordings from coral reefs
INDUSTRY-LEADING ACCURACY
We achieve an impressive 96-99% accuracy, ahead of the industry-leading 92% by University of Exeter research scientists.
LIGHTNING-FAST EFFICIENCY
Our AI model is more than 5 times more efficient compared to existing alternatives such as manual scuba-diver surveys, capable of being run in real-time.
HARDWARE ENGINEERING
Remotely-controlled
Underwater Vehicle (ROV)
Our ROV uses 4 custom-designed propellers, and processes sound recordings with a hydrophone and microcomputer. It autonomously navigates coral reefs using its FPV-camera and transmits video signal back.
Additionally, it has sensors onboard to record data (gyroscope, barometer, thermometer, ammeter/voltmeter), and hovers over coral reefs to collect sound recordings and perform diagnosis using our core ReefSound AI. It also runs our custom object-detection AI, with YOLOv5 architecture.
HARDWARE ENGINEERING
ReefBox
ReefBox is manufactured with PMMA injection-moulded casing and a clear polycarbonate lid, gasket-mounted to the top for easy access. Our waterproof AUSOMS-mini hydrophone is routed through the box and secured via 3D-printed supports.
