Amazon Facial Recognition Technology Tested at UK Train Stations

Rail passengers in the U.K. have been subjected to facial recognition scans by software from Amazon, as major network operators conducted extensive tests. 

Eight of the U.K.'s busiest train stations have been testing Amazon's Rekognition software over the past two years, according to documents obtained by Wired.

Network Rail, which manages the country’s rail infrastructure, indicated that initial tests received positive feedback for enhancing security measures.

Stations including London’s Waterloo and Euston, Manchester Piccadilly, Glasgow Central and Reading, used the software to augment existing CCTV cameras to alert staff of potential safety incidents. The railway operators used Meraki smart camera from Cisco and AWS’ Rekognition image recognition software to identify and monitor passengers.

The cameras reportedly contributed significantly to detecting trespassing incidents on tracks and preventing bicycle thefts. At Reading railway station, for example, an anti-climb barrier was installed after the cameras captured evidence of multiple incidents.

According to the documents, the facial recognition technology could accurately detect passengers as well as luggage, pushchairs and bikes.

However, Network Rail’s summary of the tests detailed that the AI software could be used to analyze people’s emotions. 

Related:Amazon Web Services extends ban on facial recognition sales to police

Network Rail suggested customer emotion data metrics “could be used to measure satisfaction” and could also be used to maximize advertising and retail revenue.

However, the rail operator found the solution only worked if the cameras had a “very clear view of people’s faces” and acknowledged that such uses are “viewed with more caution.”

The facial recognition system would kick in when passengers crossed a “virtual tripwire,” for example, near payment gates to detect fare evaders. The rail operator said the main concern of the deployment was “not fare evasion but the risk of injury.” 

The trial recommended future deployments of such systems be connected to a private, dedicated access point or wired connection, avoiding public Wi-Fi.

Beyond facial recognition, Network Rail used the AI-augmented cameras to keep track of passengers for crowd monitoring on platforms and concourses. The data obtained would enable rail stations to adopt a data-driven approach to reactively manage crowd levels, the summary suggested.

A Network Rail spokesperson told AI Business the agency takes the security of the rail network “extremely seriously” and uses a range of advanced technologies across its stations to protect passengers, colleagues and railway infrastructure from crime and other threats.” 

Related:Researchers blur faces in seminal ImageNet dataset; object recognition algorithms maintain accuracy

“When we deploy technology, we work with the police and security services to ensure that we’re taking proportionate action and we always comply with the relevant legislation regarding the use of surveillance technologies.” 

In addition to camera analytics, the rail operator also explored using low-cost IoT sensors to monitor temperature, humidity and air quality to remove the need for manual checks by human staff.

AWS has barred police forces from using its facial recognition software since 2021.