Parking information system with artificial neural network

Abstract

Intelligent parking systems are necessary for deploying smart cities. For safe and comfortable parking in the parking places, it is necessary to install an information parking system that can show the current number of free and possibly also occupied parking places. Private parking places can be limited to their owners and the access can be protected by an arm barrier or retractable post. For vehicle presence sensing or vehicle transition sensing we can use various types of sensors. Every sensing technology has its own limitations or advantages, and it does not exist one universal solution for all possible situations. This article presents and describes the parking information system (PIS) based on real-time vehicle video detection with a safety IP camera and an artificial neural network. The proposed PIS consists of trained Yolo v5 neural networks trained to detect cars, vans, and trucks. To prevent false counts many control mechanisms were implemented (Zheng, 2015)