Skip to content Skip to footer

A closer look at how we’re using AI to elevate existing LPR technology

We are all about finding innovative ways to apply our AI to solve real-world problems and this week we are focused on License Plate Recognition (LPR). This technology is a powerful tool that can be used for various applications, including parking management and in this blog piece, we’ll be showcasing a recent case study of ours, which focused on our implementation of an LPR solution for a mid-sized shopping centre.

The Problem:

The shopping centre had three parking lots – one on the retail floor and two underground lots. The main parking lot was meant for shoppers, while tenants and employees were assigned to the underground lots. However, the client faced challenges in managing the occupancy of the main parking lot. They found that tenants and staff were parking in the main lot instead of the assigned underground lots, causing inconvenience to shoppers. To solve this problem, the client had already installed LPR-capable cameras at all parking lot entrances and exits and compiled a list of number plates belonging to its tenants and staff, along with their assigned parking lot.

The Solution:

Refraime’s team developed an LPR solution that integrated with the existing cameras and tracked the specific number plates based on the list provided by the client. The system analysed the plates to ascertain where each vehicle parked and generated alerts if it detected a car in the wrong parking lot. The alert included an image of the vehicle and its corresponding license plate.
Another additional challenge posed by the project was that of inter-connected parking lots. The entrance to one of the basement lots was situated inside the main parking lot, making it a little more tricky to monitor whether the tenants moved through to the correct lot. We resolved this issue by implementing a timer that allowed the vehicle to make its way to the correct lot without triggering an alert.

During the initial stages of this project, the partial capturing of number plates by the LPR cameras installed by the centre, proved to also require some problem solving. We decided to use the Levenshtein Distance algorithm. It calculates the minimum number of single-character edits required to change one word into another. This assisted in identifying plates based on a comfortable “distance threshold.”

Conclusion:

Our LPR solution helped the client manage their parking lots better by ensuring that tenants and staff parked in their assigned underground lots. The project’s challenges were overcome by our innovative approach, and the implemented solution was deployed successfully. LPR technology has proven to be a game-changer for parking management, and this is just one of many ways we can implement it at Refraime.