Facial recog system used by Met Police shows racial bias at low thresholds
- Reference: 1685010851
- News link: https://www.theregister.co.uk/2023/05/25/facial_recognition_system_used_by/
- Source link:
Speaking to the Science, Innovation and Technology Committee, Dr Tony Mansfield, principal research scientist, National Physical Laboratory, said the NEC-based system used by the Met, the UK’s largest police force, was prone to bias against Black individuals on a set of test data created for his investigations.
"We find that if the system is run at low thresholds and easy thresholds, that it does start showing a bias against the Black males and females combined," he told MPs.
[1]
Mansfield added that he believed the Met did not operate the system at these thresholds.
[2]
[3]
The testimony before the hearing investigating the Governance of artificial intelligence (AI) related to [4]Mansfield's report [PDF] conducted by the National Physical Laboratory for the Met and South Wales Police.
[5]Phones' facial recog tech 'fooled' by low-res 2D photo
[6]UK's GDPR replacement could wipe out oversight of live facial recognition
[7]Activists gatecrash Capita's AGM to protest GPS tracking contract
[8]Smile! UK cops reckon they've ironed out gremlins with real-time facial recog
It found that false positive identifications increase at lower face-match thresholds (0.58 and 0.56) and “start to show a statistically significant imbalance between demographics with more Black subjects having a false positive than Asian or White subjects.”
However, the system, based on NEC Neoface V4 1 using HD5 Face applied to a 178,000 Filler image watchlist, offered no false positive identifications at higher thresholds (0.64 or higher) and, therefore, no bias. At the medium threshold (between 0.60 and 0.62), it offered eight false positives but no statistically significant bias in demographics.
Lindsey Chiswick, the Met’s director of intelligence, told MPs the system had been used three times since the research was published, including in the London boroughs of Camden and Islington, and during the King's Coronation.
[9]
Across these deployments, which involved the use of six police vans in all, there were four alerts, and zero false positives. There were two arrests, and with the remaining alerts, officers decided an arrest was unnecessary, even though the identification was correct.
She said that in each deployment, the police created a separate data set for the “watchlist” based on the location and the “intelligence case” for the deployment. After each deployment, the watchlist was deleted. None of the data from the live facial recognition deployments were stored, she said.
“I completely understand public concern around [areas] like facial recognition technology and AI, and there's lots of debate going on around that. I've tried introducing facial recognition technology in as careful, proportionate and transparent way possible,” she told MPs.
[10]
Nonetheless, the introduction of facial recognition in policing has attracted criticism in the UK.
In 2017, the [11]Met was urged to cancel its planned use of facial recognition software at Notting Hill Carnival, Europe’s largest street festival. Privacy groups, including Big Brother Watch, Liberty and Privacy International called for a U-turn on the use of the tech.
Speaking to MPs this week, Chiswick said the Met had only used the technology once “in the very early days of us learning to use” it. “It wasn't a very successful deployment. We learned that the sheer number of faces, the density of the crowd, made it very difficult for the technology to operate,” she said.
In 2018, campaigners [12]Big Brother Watch found 91 per cent of people flagged up on the Met’s facial recognition system turned out to be not on the watch list. At the time, a Met spokesperson said the force did not consider these false positive matches “because additional checks and balances are in place to confirm identification following system alerts”.
It's not a great look just a week after the UK biometrics and surveillance camera commissioner Professor Fraser Sampson [13]warned the government was trying to scrap independent oversight of facial recognition just as the policing minister has made it clear he plans to "embed" it into the force.
"As an advocate of the accountable and proportionate use of new technology by the police I think this lacuna is problematic as much for the police themselves as for the communities they serve," he said. ®
Get our [14]Tech Resources
[1] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_security/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZG@GKUUpqVDfiFj3aC1-CAAAAFI&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_security/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZG@GKUUpqVDfiFj3aC1-CAAAAFI&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_security/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZG@GKUUpqVDfiFj3aC1-CAAAAFI&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[4] https://science.police.uk/site/assets/files/3396/frt-equitability-study_mar2023.pdf
[5] https://www.theregister.com/2023/05/19/2d_photograph_facial_recog/
[6] https://www.theregister.com/2023/05/19/dpib_2_surveillance_oversight/
[7] https://www.theregister.com/2023/05/12/activists_gatecrash_capitas_agm_to/
[8] https://www.theregister.com/2023/04/06/uk_police_facial_recognition/
[9] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_security/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZG@GKUUpqVDfiFj3aC1-CAAAAFI&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[10] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_security/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZG@GKUUpqVDfiFj3aC1-CAAAAFI&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[11] https://www.theregister.com/2017/08/17/concerns_over_facial_recognition_at_notting_hill_carnival/
[12] https://www.theregister.com/2018/05/15/met_police_slammed_inaccurate_facial_recognition/
[13] https://www.theregister.com/2023/05/19/dpib_2_surveillance_oversight/
[14] https://whitepapers.theregister.com/
Completion
To Arrghh is to be a Pirate?
Facial recognition test
It seems people with some African ancestry may be susceptible to misidentification ...
In the imperial treasury in Vienna, there is a portrait of the Holy Roman Emperor Franz I which, probably due to centuries of inbreeding, bears an astonishing resemblance to a current royal personage* - when we visited, SWMBO saw it straight away.
Look on Google for "Franz I schatzkammer portrait"
* But, no, we didn't check the Spanish Riding School for other possible matches.
Re: Facial recognition test
Hence the concern that the Met's facial recognition will get said personage pulled over for the outstanding offences of driving a Holy Roman Empire without due care and attention, causing it to break up on the hard shoulder.
How could it possibly go wrong?
How could this possibly go wrong? Nobody could possibly think of a scenario. For example, it would never happen that:
1. If the Met were scanning a large crowd of people for a dangerous and potentially armed suspect;
2. They were certain the suspect was present (based on intelligence), but got no matches
3. They get the operators of the facial recognition system to reduce the match accuracy until some positive result was achieved
Even if this did happen, the Met would never shoot dead an innocent unarmed black person on the basis of faulty intelligence; or injure or kill them during an arrest; or detain them in circumstances leading to their death.
It's good to know that it is completely inconceivable* that any of this could ever happen.
Also:
>Mansfield added that he believed the Met did not operate the system at these thresholds.
This is the very definition of weasel words from somebody who is supposed to be researching the system, and 'did not' != 'cannot'
(* "You keep using that word, I do not think it means what you think it means" - Inigo Montoya)
Re: How could it possibly go wrong?
It can go very wrong... Even with people! On many occasions in my life, people have completely mistaken me for someone else. I met a guy at a party that was almost my twin, and we even have the same first name. Fast forward a couple of weeks, the health club I belonged to hired a new receptionist. I walked up to the counter to sign in, and the girl half looks up and says "What the hell are you doing here?" I look at her and say I'm a member of the club. She takes a second look at me and says "OMG, I thought you were my brother," the same guy I had met a few weeks earlier. Needless to say, when I asked her out, she said no! Fast forward 15 years, I'm in Rome, and an Italian woman comes up to me on the street and starts talking to me like she knew me(my wife at the time was not pleased!). I spoke in English, and she realized her mistake, 2 minutes later we walked past my twin on the same street.
Cue the Mistake song!!
Re: How could it possibly go wrong?
It’s not often you meet your doppelgänger
https://en.wikipedia.org/wiki/Doppelgänger
Have you seen meet the fockers?
"he believed the Met did not operate the system at these thresholds"
Well of course not. It's a waste of limited computing time to do what an officer can do manually.
"Mansfield added that he believed the Met did not operate the system at these thresholds."
Silly me... I had assumed he was talking about the Metropolitan Police Service, responsible for law enforcement and the prevention of crime in Greater London, and the subject of a lot of newsprint recently because of yet more reports on institutional racism, 30 years after the Stephen Lawrence affair.
It was probably one of the key requirements. "New facial recognition system should be as good as Met officers in identifying suspects on watchlists."
going to be some issues I reckon
our (very expensive and new) video conference system can't tell black people if they sit in a shady spot and just refuse to zoom in on their faces...
it's scary to think what will happen with some of the Met Police facial recognition, especially since they obviously have never lied in the past (week) & are never never never a bit trigger happy
"at certain thresholds"
I.e. : it was turned on.
To err is to be human…
You all know the rest