Austin Caballero had been getting away with it for years. A shoplifter who targeted small, high-end shops in London’s wealthier districts, he had helped himself to more than £100,000 worth of jewellery and designer clothing over an extended period.
“He was good,” says Detective Sergeant Eliot Porritt of the UK capital’s Metropolitan Police. “I hate using that word for him, but he was well dressed and calm. He would go in and engage the staff in conversation, and as soon as their backs were turned, he’d steal stuff.
“Sometimes it wasn’t until two or three days later that they’d realise something was missing from the display. Then they’d look on CCTV and call the police. But he’d be long gone by then, so he always had the advantage.”
Caballero would probably still be getting away with it were it not for individuals such as Porritt, who is one of a team of so-called ‘super-recognisers’ who have been operating at Metropolitan Police headquarters at New Scotland Yard since May 2015 and who last year lent their help to the police in Cologne, Germany.
They sound like characters from a Marvel comic and indeed their talents are close to superhuman, because they have an uncanny ability to remember and recognise faces – even faces that are only partially revealed or highly pixelated.
So when a member of the unit saw a picture of the then unknown Caballero on the Met’s computer database of CCTV images of known suspects in the summer of 2016, he decided to check and see if he had been caught on camera before. It’s a matching process the unit calls ‘face snapping’, after the game of snap, in which players look for identical cards.
After a weekend of searching, he’d snapped ten other images on the database. Eventually, after looking at tens of thousands of images, he would end up with around 40 identifications. It was clear that Caballero was a serious repeat offender.
A media appeal was launched for further information and in due course Caballero was found and later convicted of 40 offences of theft and one of racially aggravated assault. He is currently serving three years and three months in prison.
“Some of these pictures of Caballero went back to 2012,” says Porritt, an affable character with a ready smile, who is fiercely proud of the successes of the unit. “He was probably thinking, ‘I’ve committed all these offences and no-one’s ever come to see me, so I’ve got away with it.’ But that’s all changed. We’re identifying all these prolific offenders who’ve gone under the radar for years because no-one’s ever linked up CCTV images of them.”
The beginnings of the super-recogniser unit go back to the serious civil disturbances in London during the summer of 2011. It became clear then that the Met had no systematic way of dealing with large numbers of CCTV images. So the first step was the creation of a computerised database of images that could be searched by various criteria such as ethnic appearance, clothing and hairstyle. When the database was put into use, it became clear that certain officers had an uncanny ability for recognising faces.
“I first started hearing about super-recognisers in 2011,” says Acting Police Sergeant Paul Smith, who developed and now manages the seven-strong unit, plus a network of around 140 other super-recognisers, both officers and civilian staff. “When we started using the database, it became clear that certain officers, such as Eliot, were giving repeat identifications – not just one, but four, five, six, and on a regular basis.”
Porritt had no idea he had a special talent until Smith contacted him to tell him he was on his super-recogniser list. Now 37, he joined the police force in 2008. Growing up in the leafy London suburb of Belsize Park, he’d dreamed of working in the public services and making the world a better place, and it was after a job as a civilian assisting a child-abuse investigation that he realised the police force was where he wanted to be.
“I’ve always been good with names and faces but I was never aware when I was a kid of people going, ‘Oh my god, how did you remember that?’” he says, laughing.
“It’s strange, it’s only through working at the super-recogniser unit that I realise people don’t see other people the way I do. In the past I’d be looking at two pictures and go, ‘That’s the same person,’ and someone else would say, ‘Are you sure?’ And I’d go, ‘Are you blind?!’
“We super-recognisers can remember faces we’ve seen years ago. The average person can memorise 20 to 50 per cent of faces but academics tell us we can memorise 90 per cent.”
Between one and two per cent of the population has this special skill, and scientists are baffled as to why. However, it can be scientifically tested, and all the members of the super-recogniser unit have been proven to share the ability. And it’s paying dividends.
“Since the unit started in May 2015, we’ve made more than 1800 identifications,” says Porritt. “And that’s led to more than 900 completed cases.”
Porritt and colleagues regularly attend large-scale events, where they can help to identify known offenders. “A couple of the guys were at the Notting Hill Carnival in London looking at a live feed of TV images and feeding information back,” he says. “They could tell there were two gangs next to one another, so they were able to give a warning and avert a serious disturbance.”
Ten super-recognisers were also assigned to the high-profile case of Alice Gross, a teenager who went missing in west London in August 2014 and was later found to have been murdered. Their work was crucial in finding her body, which had been concealed by her killer under logs in a river.
“The key breakthrough was when we found a tiny flicker of a head lamp that had been missed by all the officers initially viewing the CCTV images in the area,” remembers Porritt. “It was a clue that the main suspect had returned to further conceal the body. The area had already been searched, but as a result of our information there was another search and she was found. From there we built a case.”
At the beginning of 2016, Porritt and a colleague went to Cologne to help police in that city investigate a huge number of sexual assaults and thefts mainly thought to have been carried out by North African refugees during the city’s New Year’s Eve celebrations. It was the first time the unit had helped a foreign force.
“There were 1546 crimes that night, including 532 sexual assaults,” says Detective Superintendent Thomas Schulte, the German police officer leading the investigation. “It was night time, so the CCTV quality was very bad. Scotland Yard called us to offer their help. I’d heard about super-recognisers before, so I was interested.”
“We were there for two weeks,” says Porritt. “They’d already identified three officers who’d made a lot of identifications and were clearly super-recognisers, so we gave them some training. When we got there, they had pictures of ten or 20 people on the wall. When we left, the wall was full of suspects.”
“I was surprised by how successful they were,” admits Schulte, who says they are now thinking of introducing super-recogniser units in Germany. “We always think about technical solutions, but this shows that the human mind is kind of interesting.”
It’s a good point. The US military recently purchased 500 pairs of X6 ‘spy glasses’ that enable a user to match a face in real time to one on a computer database. While facial-recognition software has its uses and has become prevalent in other areas, the super-recognisers demonstrate that, for now, the human brain has its advantages.
“Almost every image that we’ve got a conviction out of would never be picked up by facial-recognition software,” says Porritt. “That relies on perfect conditions, so I don’t think it will be good enough for years. CCTV cameras are usually positioned looking downwards. Angles, lighting, pixelation, facial expressions: all these change things, and that’s why you always need that human element.”
And unlike a machine, super-recognisers never stop working, and find themselves spotting wanted criminals – not to mention celebrities – in off-duty moments.
“We’re always on the look-out,” says Porritt, who on more than one occasion has also been grateful for the fact that so few people share his ability – as when he found himself late at night waiting for a bus next to someone he had once arrested. “Luckily, he didn’t recognise me,” he laughs.
THE RISE OF FACIAL-RECOGNITION TECHNOLOGY
- Keyless doorbells
A so-called ‘smart doorbell’ named Chui can be programmed to recognise the faces of friends and family members, allowing them instant entry.
- Detecting health issues
Scientists at the University of Oxford in the UK have developed a programme that can scan faces in family snaps and detect potential genetic disorders.
- Online course supervision
KeyLemon’s ‘biometric user authentication technology’ not only confirms online students’ identities, it can also make sure they are paying attention, based on the angle of their heads and how much they blink.
- Interactive cars
Ford and Intel are collaborating on technology that will allow a car to recognise the driver and automatically adjust certain features. For instance, if a teenager is driving, a parent might want to limit the car’s speed.
- Good news for cats
The CatFi Pro feeder is able to distinguish between one cat and another, allowing owners and breeders to monitor the diet and weight of individual cats.
Are you a super-recogniser? Try the ‘Short Teaser Test’ at www.superrecognisers.com