@article{Gu2022, 
author = {Sisi Gu and Xiaofei Sun and Miao Wang and Jiaqing Yu},
title = {Research on Parking-Space Occupancy Recognition Based on MobileNet and Intelligent Parking Guidance Strategy},
year = {2022},
journal = {Journal of Highway and Transportation Research and Development (English Edition)},
volume = {16},
number = {3},
pages = {78-84},
keywords = {intelligent transport, intelligent parking guidance, MobileNet, service area, parking space occupancy recognition},
url = {https://www.sciopen.com/article/10.1061/JHTRCQ.0000835},
doi = {10.1061/JHTRCQ.0000835},
abstract = {In order to solve the problem that parking space is hard to find in expressway service area, and provide accurate parking space information and effective parking guidance, a strategy of parking space occupancy recognition based on MobileNet and intelligent parking guidance is proposed. On the basis of detailed analysis of parking status and existing problems in expressway service area, this paper puts forward the classification strategy of intelligent parking guidance in service area, and builds the technical route of parking space recognition based on AI vision. Firstly, taking the high-grade video and low-grade video in the service area as the detection data sources, the lightweight MobileNet classification model is used to analyze the parking space occupancy in real time, so as to provide accurate and reliable information of the spare and occupied parking space in the parking area; Secondly, through the three-grade guidance screen, the three-grade parking guidance of main line preview guidance, entrance total capacity guidance and parking space guidance by vehicle type is realized, so that travelers can have a comprehensive understanding of parking space occupancy in the service area in advance. In order to verify the effectiveness of intelligent parking guidance classification strategy, a highway service area in North China is selected for field verification. The results show that the recognition accuracy of parking space occupancy recognition model based on MobileNet is 98.0% under sufficient illumination during the day and 90.0% at night; During peak holidays, the total travel time of vehicles due to congestion and waiting in service areas is reduced by about 7%, which can save 20% ~ 30% of the time for finding parking spaces compared with the time for finding parking spaces in traditional service areas. Therefore, the proposed parking space occupancy recognition based on MobileNet and intelligent parking guidance strategy can significantly improve the practical problems in the service area, meet the parking demand of the public to the maximum extent, relieve the parking pressure during peak hours, and improve the traffic capacity of the service area.}
}