Face Detection-Based System to Sense Pedestrians at High Risk of Collision

In this study, we propose a novel system that assesses pedestrians at high risk of collision and accordingly warns the driver of the vehicle. Numerous warning systems to prevent contact accidents between pedestrians and vehicles have recently been proposed. One desideratum of warning systems that prevalent systems lack is that they should only warn the driver about pedestrians that are at high risk of collision with the vehicle. To develop such a system, we focus on detecting pedestrian's faces. Pedestrians not looking at the vehicle are then determined to be at high risk of collision based on the face detection feature. We verified the effectiveness of our recognition system using video sequences captured by an in-vehicle camera.