London police’s use of facial recognition falls flat on its face

A “top-of-the-line” automated facial recognition (AFR) system trialled for the second year in a row at London’s Notting Hill Carnival couldn’t even tell the difference between a young woman and a balding man, according to a rights group worker invited to view it in action.

Because yes, of course they did it again: London’s Met police used controversial, inaccurate, largely unregulated automated facial recognition (AFR) technology to spot troublemakers. And once again, it did more harm than good.

Last year, it proved useless. This year, it proved worse than useless: it blew up in their faces, with 35 false matches and one wrongful arrest of somebody erroneously tagged as being wanted on a warrant for a rioting offense.

London police’s use of facial recognition falls flat on its face

Wow, talk about a failed system. The police even declared it a "success"? REALLY?!?!?

n spite of its lack of success, the Met’s project leads viewed the weekend not as a failure, but as a “resounding success,” Carlos said, because it had come up with one, solitary successful match.

Even that was skewered by sloppy record-keeping that got an individual wrongfully arrested: the AFR was accurate, but the person had already been processed by the justice system and was erroneously included on the suspect database.

So, surely the American version by the FBI (that bastion of legal integrity) has a higher success rate since they have a much larger budget I am sure! Not so much it seems.......

Studies bear out the claim that AFR is an inherently racist technology. One reason is that black faces are over-represented in face databases to begin with, at least in the US: according to a study from Georgetown University’s Center for Privacy and Technology, in certain states, black Americans are arrested up to three times their representation in the population. A demographic’s over-representation in the database means that whatever error rate accrues to a facial recognition technology will be multiplied for that demographic.

Beyond that over-representation, facial recognition algorithms themselves have been found to be less accurate at identifying black faces.

During a recent, scathing US House oversight committee hearing on the FBI’s use of the technology, it emerged that 80% of the people in the FBI database don’t have any sort of arrest record. Yet the system’s recognition algorithm inaccurately identifies them during criminal searches 15% of the time, with black women most often being misidentified.

After these repeated failings I would guess the programs would be deep-sixed. No, they are doubling down now....and plans to continue it's use on a larger scale are likely in the works. After all, they need to protect their jobs in deciding to spend so much money on a failed technology. Which as I see it, they bought a bad bag of apples, and are determined to eat them even if people get sick and die in the process.

They tested that GeStaPo/StaSi/Big Brother stuff in Berlin some days ago too.
Guess what, "our protectors" didn´t tell us how "good" the system worked.

Now i know why!

a reply to: Krakatoa

I'm pretty sure the US military have the more advanced version just only they kept it among themselves.

www.facefirst.com... seems to work for casinos but they may be throwing more money at it and use more cameras?

gnvqol.org...

Bio metric Facial Recognition To begin with, there are two forms of bio metric facial recognition today. The simplest systems are “dumb” – in the sense that all they do is take snapshots of everyone entering a premise or sitting at a table or using a machine. “Smart” systems go one extra step. They actively read the snapshots and compare them to a database that pro-actively alerts security. Even if you have been banned, on the dumber systems you can still sit and play – as long as you don’t make large bets or cause a big scene. If you do make a scene, their “dumb” systems may send over a security person undercover to take a high res picture of you and compare it to their database using a smarter system. If they have smart systems in place – you are screwed. Some state of the art systems espouse patents they gained from non-obvious designs. What do you think?

and NORA seems to be one of the more successful versions

NORA – Non Obvious Relationship Awareness NORA is a system of both love and hate. NORA is the system that the Department of Homeland Security started using to identify terrorists and terrorism links after 9/11. Yeah – it can help find and defeat terrorists. But Blah – because with little more than a relationship it gives the government an excuse to trample on your liberties and rights. If you have facial recognition combined with NORA, it can show you how two people sitting at a table together might be related – even if it is because they were frat brothers 20 years ago and 5 states away. It can also show that the dealer is a distant cousin of said player who started hitting it big. NORA most recently goes by the name IBM Relationship Resolution, we personally liked NORA better. But you get the idea. NORA is designed to recognize in seconds relationships that humans could never know at first glance. NORA can also grab criminal records and local arrest records.

few more links this one compares comparable softwear www.kairos.com...

www-03.ibm.com... IBM is throwing in voice recognition as well

a reply to: RalagaNarHallas

Casinos have been using the technology for several years. I remember a buddy of mine talking about it as early as the beginning of the 2000s. With a) more money to throw at it and b) nearly two decades of tweaking it, I don't find it surprising at all that they're miles ahead of the versions mentioned in the OP.

a reply to: RalagaNarHallas

Well, if there are proven more accurate systems, then the question must be why these others were chosen and money spent (and still being spent) upon them following repeated public failures.

Can you say conspiracy to defraud the public?

Could it be a case of lobbyists influencing the purchasing of expensive systems, while the decision makers get rich from the kickbacks and junkets that usually go along with that type of "deal"?

a reply to: Krakatoa

I would never rely on that failed system for anything, it's all bollocks anyways, and a waste of taxpayers money.

a reply to: Krakatoa

The cameras aren't there to catch criminals...

originally posted by: Shamrock6
a reply to: RalagaNarHallas

Casinos have been using the technology for several years. I remember a buddy of mine talking about it as early as the beginning of the 2000s. With a) more money to throw at it and b) nearly two decades of tweaking it, I don't find it surprising at all that they're miles ahead of the versions mentioned in the OP.

Training is important too. There are parameters, if the operator isn't inputting the right data the computer will be skewed.

Open the Pod Bay doors please, HAL.

originally posted by: RalagaNarHallas
www.facefirst.com... seems to work for casinos but they may be throwing more money at it and use more cameras?

gnvqol.org...

Bio metric Facial Recognition To begin with, there are two forms of bio metric facial recognition today. The simplest systems are “dumb” – in the sense that all they do is take snapshots of everyone entering a premise or sitting at a table or using a machine. “Smart” systems go one extra step. They actively read the snapshots and compare them to a database that pro-actively alerts security. Even if you have been banned, on the dumber systems you can still sit and play – as long as you don’t make large bets or cause a big scene. If you do make a scene, their “dumb” systems may send over a security person undercover to take a high res picture of you and compare it to their database using a smarter system. If they have smart systems in place – you are screwed. Some state of the art systems espouse patents they gained from non-obvious designs. What do you think?

and NORA seems to be one of the more successful versions

NORA – Non Obvious Relationship Awareness NORA is a system of both love and hate. NORA is the system that the Department of Homeland Security started using to identify terrorists and terrorism links after 9/11. Yeah – it can help find and defeat terrorists. But Blah – because with little more than a relationship it gives the government an excuse to trample on your liberties and rights. If you have facial recognition combined with NORA, it can show you how two people sitting at a table together might be related – even if it is because they were frat brothers 20 years ago and 5 states away. It can also show that the dealer is a distant cousin of said player who started hitting it big. NORA most recently goes by the name IBM Relationship Resolution, we personally liked NORA better. But you get the idea. NORA is designed to recognize in seconds relationships that humans could never know at first glance. NORA can also grab criminal records and local arrest records.

few more links this one compares comparable softwear www.kairos.com...

www-03.ibm.com... IBM is throwing in voice recognition as well

Smaller data set. How many people go to a casino on a regular basis? Can't be more than a few hundred.

The USA tried this 15 years ago:

usatoday30.usatoday.com...

Face recognition software works on the W shaped area of the face defined by the eyebrows, eyes, nose and mouth.
Obscure or modify those and it fails.

a reply to: Krakatoa

Really? So that stuff on the tv shows where they identify the perp by way of a CCTV shot is all bunk? What the hell point is there in having the damn cameras in the first place? So there's all these cameras down town and they've got some morbidly obese contractor dude watching six screens as he finishes off another cola and bag of cheetos and watches as some guys mug a woman and run off with her purse and get away without the cops being able to identify the perps?

Good grief..............what a friggin' joke that is.

originally posted by: TonyS
a reply to: Krakatoa

Really? So that stuff on the tv shows where they identify the perp by way of a CCTV shot is all bunk? What the hell point is there in having the damn cameras in the first place? So there's all these cameras down town and they've got some morbidly obese contractor dude watching six screens as he finishes off another cola and bag of cheetos and watches as some guys mug a woman and run off with her purse and get away without the cops being able to identify the perps?

Good grief..............what a friggin' joke that is.

Yes, real-life does not follow a Hollywood script. Welcome to the real world, where the cameras on the poles are watched by a human, and likely someone with little training and a lot of time to sit around waiting for something to happen.

The resolution of a high-resolution video camera is 1280x1024 at best. Now you have to factor in the field of view. This is the angle that the camera sees. Humans have a field of view of 160 degrees. Stretch your hands out in front of you and then move them to the sides until you can just about see them - that's field of view. Those 1280 pixels have to be shared evenly across this angle. As distance doubles, the amount of space covered by each camera pixel doubles. At 1 meter distance, its 0.1cm/pixel. 10 meters = 1cm/pixel, 100 meters = 10cm/pixel. A human face is 20cm across. HD camera might do 50x zoom, so that reduces the field of view and improves resolution for 100 meters. But then you have to worry about street lighting, blurriness, fog, rain,dazzled from car headlights.

One of my workplaces used to have an employee car park that had a public right of way going right along it. Employees would have their cars broken into and burgled while they were working. The CCTV was as useful as a chocolate kettle because they wouldn't clearly identify anyone beyond 10 meters.

a reply to: stormcell

I know exactly what you mean about crappy CCTV. A lot of commercial users simply check a box to say they do CCTV, while using the cheapest possible option, recording at a low frame rate and saving small compressed images which, when expanded show a pixelated mess, well below even old school VHS system capabilities.

I preferred using newer HD systems which I had installed at many locations and it blew those who viewed it away, even colleagues from across the pond, who then asked why they didn't have it.

A lot of times it''s simply down to cost, although the extra storage required for providing higher frame rates and image sizes is cheap enough these days.

Failed now, but whats the world gonna look like in 40 years. eesh

a reply to: Krakatoa

Perhaps they fooled you all me thinks.....

To use the system when others don't want them to, why not simply pretend it doesn't work by temporarily tweaking the alorithms, put it on show and the focus of attention shifts to something that presents more of a threat leaving free, functional use of the system in future.

Classic.

originally posted by: nerbot
a reply to: Krakatoa

Perhaps they fooled you all me thinks.....

To use the system when others don't want them to, why not simply pretend it doesn't work by temporarily tweaking the alorithms, put it on show and the focus of attention shifts to something that presents more of a threat leaving free, functional use of the system in future.

Classic.

They are not that intelligent. You give them too much credit, As I said real life is not like a Hollywood movie plot. Far from it....

originally posted by: Krakatoa

A “top-of-the-line” automated facial recognition (AFR) system trialled for the second year in a row at London’s Notting Hill Carnival couldn’t even tell the difference between a young woman and a balding man, according to a rights group worker invited to view it in action.

Because yes, of course they did it again: London’s Met police used controversial, inaccurate, largely unregulated automated facial recognition (AFR) technology to spot troublemakers. And once again, it did more harm than good.

Last year, it proved useless. This year, it proved worse than useless: it blew up in their faces, with 35 false matches and one wrongful arrest of somebody erroneously tagged as being wanted on a warrant for a rioting offense.

London police’s use of facial recognition falls flat on its face

Wow, talk about a failed system. The police even declared it a "success"? REALLY?!?!?

n spite of its lack of success, the Met’s project leads viewed the weekend not as a failure, but as a “resounding success,” Carlos said, because it had come up with one, solitary successful match.

Even that was skewered by sloppy record-keeping that got an individual wrongfully arrested: the AFR was accurate, but the person had already been processed by the justice system and was erroneously included on the suspect database.

So, surely the American version by the FBI (that bastion of legal integrity) has a higher success rate since they have a much larger budget I am sure! Not so much it seems.......

Studies bear out the claim that AFR is an inherently racist technology. One reason is that black faces are over-represented in face databases to begin with, at least in the US: according to a study from Georgetown University’s Center for Privacy and Technology, in certain states, black Americans are arrested up to three times their representation in the population. A demographic’s over-representation in the database means that whatever error rate accrues to a facial recognition technology will be multiplied for that demographic.

Beyond that over-representation, facial recognition algorithms themselves have been found to be less accurate at identifying black faces.

During a recent, scathing US House oversight committee hearing on the FBI’s use of the technology, it emerged that 80% of the people in the FBI database don’t have any sort of arrest record. Yet the system’s recognition algorithm inaccurately identifies them during criminal searches 15% of the time, with black women most often being misidentified.

After these repeated failings I would guess the programs would be deep-sixed. No, they are doubling down now....and plans to continue it's use on a larger scale are likely in the works. After all, they need to protect their jobs in deciding to spend so much money on a failed technology. Which as I see it, they bought a bad bag of apples, and are determined to eat them even if people get sick and die in the process.

Dont put too much weight on this. This just early stuff, given time it will be perfected, rolled out and become a primary method, along with cashless financial transaction history to trace the movements and actions of the masses.

When are people going to learn, they are regard by the govt as the enemy of the govt. thats why they are keeping us under surivileance.

Dr Day said in his lecture to another group of doctors entitled "every thing is in place, no one can stop us now," among the many other thigns hes aid, that everything has two uses.

originally posted by: starwarsisreal
a reply to: Krakatoa

I'm pretty sure the US military have the more advanced version just only they kept it among themselves.

No way, if that was a working system they would be selling it and making coin, pretty obvious it is bunk.

Sounds like due to being from a smaller database, as the FBI has the issue as well. Check into Facebook's facial recognition which am sure many more will want access to:

Facebook, according to the company, is able to accurately identify a person 98 percent of the time.

Sou rce