They have an exceptional memory for faces and an ability to identify well above average. At a time when the world is debating artificial intelligence and facial recognition technologies, the English police are relying on a few “super-recognizer” officers.
Only 1% of the population has this innate ability, “this superpower”, according to Tina Wallace, a Thames Valley Police officer who covers a large area in west London.
His team, which began recruiting these special agents in 2017, now has around 20 “super-scouts”, including Alex Thorburn, who has worked in the police for 17 years.
“I’ve always been good at recognizing faces. So when I was offered to take tests, I took the plunge,” says the latter. The results confirmed his suitability.
After a week of training, she did one last exercise, in a shopping center. “I was shown photos, dating from 10 to 30 years ago, of ten people. They had to be found in the crowd. I succeeded, even though they had changed a lot.”
The team works both behind screens, using video surveillance images, and in the field.
Alex Thorburn was thus placed in the crowd gathered at Windsor Castle, west London, for the coronation of Charles III in May.
“We were shown many photos of people, obsessed with the royal family. Our role was to see if they were present in order to avoid any problems. Fortunately, everything went well”.
“It’s a cheap and effective way to fight crime,” boasts Mike Neville, a retired Scotland Yard investigator, where he set up the first team of “super-scouts”. He then created the association Super Recognisers International.
One of the first “major successes” of these extraordinary police officers dates back to 2011, at the time of the London riots. There were a total of 200,000 hours of CCTV footage. “Twenty officers identified 600 rioters,” said Josh Davis, a professor of applied psychology at the University of Greenwich, who is a member of the Mike Neville association.
A police officer, who was an expert in gangs, recognized by analyzing the images 180 offenders, including some he had never seen in person and who had part of their face hidden.
The Thames Valley Police now deploy their “super-scouts” on certain evenings outside bars and nightclubs to spot known perpetrators of sexual assault. “Undercover agents are on the ground or behind video surveillance. As soon as they spot inappropriate behavior, uniformed police are sent to the scene,” says Tina Wallace.
In three years, 520 people have been arrested. “Two out of five men we arrested had previous convictions for rape or sexual assault,” adds the 50-year-old woman, 26 of whom are in the police.
But in the age of artificial intelligence (AI) and facial recognition technologies, aren’t these human skills outdated?
“There is no competition. They can be used together”, as the London police do, assures Mike Neville. “AI works well with good quality images. Humans are better at low quality, when the face is tilted or partially covered by sunglasses or a mask.”
“Additionally, under UK and EU law, AI identifications must be verified by a human before an arrest is made,” he notes. “Most people in a democracy would be happier if people made the decision to detain someone, rather than a computer.”
Demand for “super-recognizers” continues to grow, including police in Germany and Australia, according to Mike Neville.
Professor Josh Davis has posted a basic test in which you can earn a maximum of 14 points: “If you score less than 10 or 12, you are not going to become a super-recognizer. But if you reach 14, please contact me!”
09/01/2023 16:36:10 – Sulhamstead (Royaume-Uni) (AFP) – © 2023 AFP
