Abstract

Parking and searching for parking are becoming an increased concern. Understanding parking behaviours and patterns by local authorities could help in providing better parking management that would tackle those issues.

This paper aimed to present an artefact and further suggestions that could be used to optimise or improve the council’s pay-and-display parking with the use of computer vision and predictive analysis, as well as enhancing other parking space types. Parking video footage and multiple datasets have been used and created, and the artefact has been subject to usability testing.

The results suggest that the parking locations are not used to their full capacity, the weather and time/day information influences the behaviours, but no correlation was found for the school holidays presence. Discrepancies in the number of valid parking tickets and CCTV/video count were found. Popularity and non-popularity times, days and charging points were discovered for electric vehicles, as well as time differences between charging and just being plugged in. Moreover, the usability testing showed that the proposed artefact’s best features are CCTV/video occupancy check or the live count, the up to 2-hour vacancy predictions, the map view and the individual page locations. The obtained results suggest that the proposed artefact could serve the local authorities as a tool that could be used to manage their pay-and-display parking and to use it for future policies, funding, workforce distribution, revenue improvements and to tackle the parking search issue.

Use of Computer Vision and Predictive Analysis for Parking Spaces