Leveraging the latest computer vision and deep learning technology, Hazen.ai’s 3D analytics engine senses each vehicle, pedestrian, or cyclist as an object on a three-dimensional environment map. The proprietary patent-pending technology can provide live real-time 3D sensing capability at the edge, by processing a video feed from an off-the-shelf IP video camera.
“3D video analytics gives us the best of both worlds,” says Muhammad Amin, CTO of Hazen.ai. “We have engineered it to give radar-like capability while maintaining the unique advantages of vehicle trajectory analytics. It is sort of like the 3D map view you see in a Tesla car, but from roadside-mounted video cameras.”
Hazen.ai’s team is calling the new technology “a digital twin of your traffic camera feed”, and indeed it does feel exceptional and different when viewed in their demo. The engine can provide a view of the traffic feed from an angle of your choice in real-time. Without the need of any additional sensor, the engine provides a live view of the traffic that ensures accurate positioning of vehicles on the road and relative to each other.
“We believe this is a revolutionary step in traffic sensing,” says Sohaib Khan, CEO of Hazen.ai. “There is so much more that you can do with 3D sensing. It provides an accurate understanding of the vehicle orientation, its position relative to other vehicles, its ground footprint, dimensions and more. All this is not possible with just 2D rectangular bounding boxes that most deep learning engines provide.”