BUILDING A VEHICLE DETECTION SYSTEM BY USING DEEP LEARNING MODEL YOLO3

Abstract

Vehicle detection is a computer vision problem that has many useful applications in automatic driving systems, transportation vehicles management and traffic flow at important intersections and roads. There are a lot of approaches for this problem, from background subtraction ones to modern deep learning methods. In this paper, authors focus on applying deep learning model YOLO3 (You Only Look Once version 3) to deal with the problem. A demo system is built based on Darknet-53 and tested with self-collected data. The obtained results show that our system gains high accuracy and is viable in real life situations.