Live facial recognition technology has been employed by London’s Metropolitan Police in a second major shopping location.
Oxford Street, the UK’s busiest shopping destination, yesterday became the latest target of the Met’s live facial recognition CCTV camera, scanning anyone who walks past to find “people who are wanted for violent and other serious crimes”.
This controversial technology was deployed last week outside Westfield Stratford shopping centre and received a similarly hostile reception from privacy campaign groups.
Facial recognition cameras spotted right now on top of a van by Oxford Circus: pic.twitter.com/xPTAMkVAX5
— Big Brother Watch (@BigBrotherWatch) February 20, 2020
The Met announced plans to use the technology on Twitter stating that it was would deploy the camera van at “key locations” in Westminster scanning peoples’ faces and matching them against a database of 5000 people wanted for serious crimes.
Campaign group Big Brother Watch (BBW), who picketed the van’s previous deployment outside Westfield Stratford last week, responded on the social media platform stating that the Met’s tweet “is a lie”.
READ MORE: UK’s first live facial recognition CCTV deployed at Westfield Stratford shopping centre
“It’s not being used only for serious crime & can include innocent people,” it added.
“Police try to exploit our fears to take our freedoms. We won’t allow it.”
According to BBW 93 per cent of those stopped during the Met’s previous 10 public trials were wrongly identified.
A further study published last month by a surveillance expert from Essex University Professor Pete Fussey, who conducted the only independent review of the Met’s trials, found that just 19 per cent of people identified were done so accurately.
The Met has contested these figures, stating that their system correctly identified 70 per cent of people who walked past the camera, with an error rate of just 1 in 1000.
Click here to sign up to Charged free daily email newsletter
1 Comment. Leave new
How does “correctly identified 70%”, which implies incorrectly identified 30%, or an error rate of 3 in 10, square with the claimed “error rate of just 1 in 1000”?