Powered by patent protected processes and leveraging a custom network trained on an extensive archive of coastal imagery, Surfzone AI returns significantly better detection results in the ocean environment than a ‘best in class’ generic object detection model.
Our model understands a wide variety of coastal features at the pixel level, including waves (whitewater, unbroken face, plunging lip, spray), beach, crowds, and objects. This allows for detailed data to be derived on any aspect of the surf zone environment.
Combining computer vision designed for small object detection with a PTZ camera, we can generate full attendance monitoring from a single camera location. We can fully reconstruct the scene and map position.
Our camera system requires minimal setup and configuration. With cameras capable of mounting almost anywhere there's connectivity, our software automatically reconfigures in the event a camera moves intentionally or inadvertently.
We have broad patent protection in Australia and the United States for our approach to deriving recreational and environmental data in the surf zone, with similar patents pending in Europe.
Risk models and long term analysis are highly vulnerable to false negative detections. If the camera is obscured by fog, rain, dirt or obstruction our model detects that and flags the data as unreliable.
Our system deals with occlusion (people obscured by waves or each other) with a date led statistical averaging through time coupled with high frame rate analysis.
%20On%20LIght.png)
