Object Detection and Classification for ADAS and Autonomous Driving

Abstract

A reliable and robust perception system of the real world is a necessary for an autonomous vehicle and the Advanced Driver Assistance Systems. Obstacles detection and classification are the main pillar for the correct understanding of the dynamic world. The system can provide a reconstruction of the dynamic world surrounding the vehicle, proving to be able to help the driver in the assessment of critical situations. In particular, the developed algorithm provides a stable, robust and reliable detection, classification and tracking of the multiple targets coming from different cameras.