Development of an Automated Electronic Toll Collection System Based on Deep Learning

Road toll collection is an indispensable activity of a country, contributing to offset construction costs as well as having more capital to improve road construction. Therefore, the construction of toll stations is mandatory. In order to achieve the highest satisfaction of traffic participants and transport businesses, reduce traffic congestion at toll stations, minimize environmental pollution, we develop a real-time automatic toll collection systems based on deep learning. Our system consists some of advanced technology methods includes: (i) The development of YOLOv5 to detect the car license plate; (ii) The implementation of PaddleOCR to recognize the number license plate; (iii) The development of Web Server to manage the system; (iv) the development of lab-scale express highway used Vietnam Electrical Toll Collection (VETC) to ensure transparency in the toll collection.