The method for providing of parking location guidance service in a smart parking lot

Abstract

We proposed a parking location guidance service method using deep learning model in a smart parking lot. License plate image detection and character recognition are required to guide the parking location, and both of these steps use machine learning methods. In particular, the edge system and the server are mounted with a learning model so as not to generate a lot of traffic by transmitting all vehicle images to the server, and the edge system cut only the license plate image using the vehicle license plate detection learning model and transmits it to the server. In the server mounted with the recognition learning model, only license plate images are collected to recognize license plate characters. Rather than performing both license plate image detection and license plate character recognition in the server, it will be much more efficient to cut the license plate image in the edge system and transmit only image information necessary for character recognition to the server as it can reduce the traffic load.