Advanced safety applications and autonomous driving rely on 3D perception. Bird’s-eye-view (BEV) is an increasingly popular solution to create a highly accurate 360° view around the ego vehicle.

In this example, a footage from 6 cameras with 2MP resolution was used to run a perception neural network at 25 FPS.

Processing the data from all 6 cameras into a 3D surround view requires high compute power. The Hailo AI accelerators perform real-time inferencing of advanced AI pipelines with high accuracy and at an affordable cost, enabling BEV for mass-market vehicles.