Vehicle Detection and Counting Systems using YOLO Deep Learning Framework

數位影像處理 Digital Image Processing Final Task | 2018 | Course ID EE5412701
Aji Setyoko,  Hendrik Tampubolon 
National Taiwan University of Science and Technology, Taiwan
Proposed Method (Algorithm Flow)

Abstract


The real time vehicle count is necessary to know the real-time traffic monitoring as the additional information of Global Positioning System data. A simple tracking method is combine with Vehicle detection based on real-time object detection to produce very robust, very light, and high accuration real-time vehicle counter. Region of Interest and High-Pass Filter method is also employed to give a high reliability of the model. By tested the model on real traffic monitoring video shown the model get 99% accuracy. 

[Code] [Demo]

Presentatioin

Video Demo