DEVELOPING A SYSTEM FOR VEHICLE TRACKING AND SPEED ESTIMATION

Abstract

Speed has always a biggest risk factor in traffic. In 2022, speeding was a contributing factor in 29% of all traffic fatalities. Radar speed camera is a way to measure vehicles speed, but it is expensive to install and complex installation. A way to measure vehicles speed through camera would solve this problem since the number of traffic camera has been increasing recent years. Camera can monitor multiple aspects of traffic such as vehicle classification, lane tracking and speed estimation. With the computer’s computational power has increased, we can now monitor traffic with large number of vehicles at the same time in real-time. This will greatly reduce the cost for radar speed camera and potentially reduce accidents on road. In this thesis, we perform experiments with YOLOv8 and YOLOv10 for vehicles tracking. We use the Vehicle_Detection Computer Vision Project dataset for classify vehicles and BrnoCompSpeed dataset for the speed estimation. The experiment result show a good accuracy and this research have the possibility to implement on real life application.