Pre-collision Assist with Pedestrian Detection Using AI

Abstract

This paper presents a project on a driver assist system which uses the concept of computer vision and a deep learning to detect and identify any pedestrians who suddenly step into oncoming traffic so as to provide an alert or warning to the driver of the vehicle in case of a collision. Since pedestrian-related accidents are increasing every year, with the development of artificial intelligence, machine learning, and deep learning-based pedestrian detection methods has greatly improved the accuracy of pedestrian detection and object tracking becomes much easier. In this project, a deep learning algorithm known as the YOLO algorithm which uses a joint training algorithm is used which helps in the detection and classification of objects. Therefore, with the use of both cameras and the concept of artificial intelligence an efficient pedestrian scanning and identification method can be developed.