Using machine vision to fight crime: Real-time license plate detection and notification

Cameras have been reading vehicle license plates for things like issuing speeding tickets, deducting tolls and conducting border surveillance.

What if we could use cameras to read license plates to identify criminal suspects and notify the nearest police in real time? This is what we set out to do at DXC Labs using machine vision.

Machine vision is a growing market, “expected to reach USD 15.46 billion by the end of 2022 with 8.18% CAGR during forecast period 2017-2022.” (Source: Wikipedia)

Key differentiators

In our prototype, we are using optical character recognition (OCR), a type of machine vision, to read and digitize license plate information. We can read Australian, European Union and United States plates with 85 percent accuracy or better.

Our key differentiators are a reusable processing framework in the cloud and the ability to do real-time processing remotely.

How it works

While most cameras have built-in “detect and read” license plate processing, our prototype uses an analytics framework built on Microsoft Azure with Microsoft’s Cortana Intelligence Suite. This reusable framework handles the image recognition logic (leveraging openALPR) and the relevant analytical post-processing of ingested images or video streams.

The processing logic uses Azure Stream Analytics to verify, in real-time, if the read plate is listed as a vehicle involved in a previous criminal incident in a DMV or police database. If the plate is listed, our system immediately sends an audio message that can be integrated with the local police department’s (RF) transmission system.

Vision analytics architecture for reading license plates

Vision analytics architecture for reading license plates and notifying law enforcement in real time of crime-listed plates. Click image to expand.

Currently, we do not know of a system that provides real-time alerts like this to the nearest police on duty.

Other applications

We have also designed a locator system, where one can search for a specific license plate on a certain date to check the route the vehicle passed through. This system, designed in Microsoft’s Power BI, can be used for investigation, as in claims analysis and other relevant verification. For example, when an automobile claim is being adjudicated, this information can help verify whether the specific vehicle was involved, along with date-time-location coordinates and a vehicle image for authentication.

Using vision analytics to review the path of a crime-listed vehicle. Click image to expand.

A wide range of post-processing analytics and applications exist. For example, sudden changes in the volume of plates read at a certain time can be used to detect traffic patterns and provide alerts about traffic delays, potential road repair and other situations.

Transportation services companies, insurance companies, utilities and manufacturers should think about how they can use machine vision to address business challenges or open up new opportunities. Don’t be short-sighted — use your imagination!

Sukanya Kuppuswamy focuses on machine learning and advanced analytics in DXC Labs in India. A data enthusiast, her experience includes consulting and developing technology solutions for multiple business situations in the insurance industry. @sukanyaswamy




Using data stories to accelerate machine learning solutions

V is for visionables

Artificial intelligence in travel and transportation: How to take care of the fleet, the business and the passenger




Speak Your Mind


This site uses Akismet to reduce spam. Learn how your comment data is processed.