Rekor also uses the image recognition platform to help police departments manage "hot lists" of stolen and uninsured cars.
Rekor Systems, a new artificial intelligence company, is modernizing automatic license plate recognition to do everything from charge tolls to customize drive-thru menus to spot stolen cars.
Rekor's newest partner is Mastercard in a project to help fast-food companies increase customer loyalty and make the just-in-time food prep process even more efficient.
Mastercard is launching pilot projects with Circle K, Dunkin' and White Castle that uses Rekor's licence plate recognition technology and SoundHound's voice-recognition software. This AI-powered drive-thru system uses vehicle recognition and voice ordering to personalize the ordering experience for repeat customers and to create more touchless order options.
Rekor's ALPR system will watch for a customer's car to enter the parking lot and alert the restaurant when the customer arrives. This could trigger a menu display that features the customer's usual order or alert the food prep workers to prepare an order that has already been placed.
Rekor CEO Robert Berman said that he had been working on a similar project with a fast-food restaurant chain for the last year.
"There's been three or four other companies that have tried to do this and approached it from different angles but Mastercard is Mastercard and they have access to these stores at a corporate level," he said.
Rekor will provide a camera for each store in the pilot project that will use the ALPR recognition software. The platform works with existing IP cameras and the image analysis can be done at the edge as well as the cloud.
SEE: Natural language processing: A cheat sheet (TechRepublic)
Rekor's licence plate recognition platform collects more data than cameras that use optical character recognition (OCR) technology to record license plate numbers. Rekor's IP cameras feed data to the cloud where algorithms can identify the make and model of the car as well as the color. This advanced recognition is due to the ability to identify colors, a capability that OCR software traditionally has not had. Rekor's Chief Science Officer Matt Hill built this capability at OpenALPR and Rekor acquired the company in March 2019.
"The algorithm reads the video stream across the whole color spectrum, which creates the opportunity for us to ask customers what they want us to identify," he said. "And it's easier to do the annotation with machine learning."
Berman said that the image analysis is so accurate in identifying the make, model, and color, that many times this information can be enough to find the vehicle without having the full plate number.
Hyundai has built an app using Rekor software to identify vehicles that have been recalled and in Brazil, a mall uses the technology to allow customers to earn loyalty points to pay for parking costs.
Tracking stolen and uninsured cars
In addition to commercial customers, Rekor works with police departments around the country to manage "hot lists" that include stolen vehicles, vehicles involved in a crime, and vehicles associated with a person the police department is tracking. ALPR systems send an alert when the camera system spots one of these plates.
Departments also use the platform to track uninsured vehicles and parking tickets. Berman said that contactless solutions can be a better approach for addressing these infractions.
"When you think about what's happening today, the less engagement that public safety officers have with people when they don't have to can be a good thing," he said.
Rekor also has a citation management platform that customers can use to send diversion notices to drivers without insurance. Transportation departments also use Rekor's cameras and software to manage tolls on bridges and roads.
Berman said that another advantage of the platform is that the same camera can be used at the same time for different activities. Police officers can use the data stream from a camera to look for cars on a hot list and a department of transportation can use the same data to do a real-time traffic study. Rekor uses a federated approach to data distribution.
"The data stays with the originator and it's up to the customer to share it," he said.
Privacy concerns about ALPR
As with most surveillance technology, there are privacy concerns around ALPR. The technology is sold as a tool for tracking bad guys but the reality is that the cameras capture all traffic that goes by and often stores the license plate data for future use.
In 2018, the Electronic Frontier Foundation analyzed records from 200 police programs that use this technology. The researchers found that a total of 173 police departments, sheriff's offices, and federal agencies in 23 states scanned 2.5 billion license plates in 2016 and 2017. On average, 99.5% of scans belonged to cars that were not connected to a crime.
The power of the platform is to be able to share this information easily and agencies were doing just that. EFF also found that each agency was directly sharing this information with an average of 160 other jurisdictions, including other police departments, university police units, airports, or Customs and Border Protection. Some ALPR software vendors maintain a database of all the information collected by the system and then sell access to this dataset to government agencies and private companies.
- IT leader's guide to deep learning (TechRepublic Premium)
- Building the bionic brain (free PDF) (TechRepublic)
- Hiring Kit: Autonomous Systems Engineer (TechRepublic Premium)
- What is AI? Everything you need to know about Artificial Intelligence (ZDNet)
- 6 ways to delete yourself from the internet (CNET)
- Artificial Intelligence: More must-read coverage (TechRepublic on Flipboard)