Beta
×

Welcome to the Slashdot Beta site -- learn more here. Use the link in the footer or click here to return to the Classic version of Slashdot.

Thank you!

Before you choose to head back to the Classic look of the site, we'd appreciate it if you share your thoughts on the Beta; your feedback is what drives our ongoing development.

Beta is different and we value you taking the time to try it out. Please take a look at the changes we've made in Beta and  learn more about it. Thanks for reading, and for making the site better!

Face Recognition Algorithm Finally Outperforms Humans

Unknown Lamer posted about 5 months ago | from the man-is-obsolete dept.

AI 68

KentuckyFC (1144503) writes "Face recognition has come a long way in recent years. In ideal lighting conditions, given the same pose, facial expression etc, it easily outperforms humans. But the real world isn't like that. People grow beards, wear make up and glasses, make strange faces and so on, which makes the task of facial recognition tricky even for humans. A well-known photo database called Labelled Faces in the Wild captures much of this variation. It consists of 13,000 face images of almost 6000 public figures collected off the web. When images of the same person are paired, humans can correctly spot matches and mismatches 97.53 per cent of the time. By comparison, face recognition algorithms have never come close to this. Now a group of computer scientists have developed a new algorithm called GaussianFace that outperforms humans in this task for the first time. The algorithm normalises each face into a 150 x 120 pixel image by transforming it based on five image landmarks: the position of both eyes, the nose and the two corners of the mouth. After being trained on a wide variety of images in advance, it can then compare faces looking for similarities. It does this with an accuracy of 98.52 per cent; the first time an algorithm has beaten human-level performance in such challenging real-world conditions. You can test yourself on some of the image pairs on the other side of the link."

cancel ×

68 comments

Sorry! There are no comments related to the filter you selected.

Marketing... (1)

Torp (199297) | about 5 months ago | (#46822171)

and/or sensationalism... a 13000 image set a whole population does not make.
It's probably just better than the existing algorithms somewhat.

Re:Marketing... (5, Interesting)

BasilBrush (643681) | about 5 months ago | (#46822653)

But Human's don't match faces based on the entire population either. Just on the faces they know.

I don't know how many people the average person can recognise, but my guess is that it will be less than 13,000.

This anthropologist seems to have worked in this area, and he puts the number of people you can recognise and put a name to as 1500. (You'll recognise more than that, but you won't have names to go with them.)
http://spectrum.ieee.org/telec... [ieee.org]

Re:Marketing... (0)

Anonymous Coward | about 5 months ago | (#46825813)

What about confirming/refuting identity between two different images of what may or may not be the same stranger or two different strangers?

One does not need to be familiar with the individual, in order to recognize an individual's face in two separate images, or assess the two images of two faces as belonging to two different people.

Re:Marketing... (0)

Anonymous Coward | about 5 months ago | (#46830957)

Maybe a trolling comment, but that's pretty obvious from studies that have been done in photo/mug shots as well as live lineups. And as weve seen from the number of innocent people imprisoned because the witness wont admit the never truly saw the person, or they say the police pressured them to ID someone.

You can keep reporting about these algorthims all you want but tech similar to MIR scan will emerge. Able to scan skull and muscle tissue and put an end to this foolishness of building an algorithm.

Re:Marketing... (1)

BasilBrush (643681) | about 5 months ago | (#46834849)

MIR scanning will tear the piercings right out of people's ears and faces.

Re:Marketing... (3, Insightful)

shadowrat (1069614) | about 5 months ago | (#46824419)

and/or sensationalism... a 13000 image set a whole population does not make. It's probably just better than the existing algorithms somewhat.

I doubt my internal facial recognition distinguishes between 13000 people. And worse, as damning as it sounds, it really starts to fail when dealing with people of certain races that i just didn't get much exposure to as a kid.

Re:Marketing... (1)

Matheus (586080) | about 5 months ago | (#46828313)

That's not really the point... this is a *specific set of known hard to match faces that is used as a measuring stick for Facial Recognition algorithms. The test is checking individual 1:1 compares NOT a 1:N search. Converting a 1:1 algo to 1:N work is a whole 'nother topic. Honestly nothing that they describe in the article is new (normalizing on facial features and fixed resolution images and Vectorizing the image) so hard to see where they made their improvements. My former employer had the best Facial algorithms in the biz up to and including the ability to distinguish identical twins given a high quality image and no data set training required BUT would be interesting to see how their algorithms hold up in comparison.

money (2)

watcher-rv4 (2712547) | about 5 months ago | (#46822179)

NSA's financing?

Re:money (1)

gatkinso (15975) | about 5 months ago | (#46822253)

NSA mission does not include IMINT or HUMINT.

Re:money (1)

fuzzyfuzzyfungus (1223518) | about 5 months ago | (#46822301)

And there are no digitized images or videos that include faces anywhere among the data they haul in?

Re:money (0)

Anonymous Coward | about 5 months ago | (#46822383)

Do you know what SIGINT is?

Re:money (1)

fuzzyfuzzyfungus (1223518) | about 5 months ago | (#46823695)

Yes, and I don't see why faces, especially if you have a method for quickly and easily algorithmically fingerprinting them, would be of any less interest than any other aspect of an intercepted signal that could be used to draw inferences about who and where it came from, where it's going and who it's intended for, draw correlations between otherwise apparently separate transmissions, and so on.

Given the relative obscurity of video compared to voice and text, I wouldn't see it replacing CDR grovelling or anything; but that's merely a question of scale rather than of scope.

Re:money (1)

donscarletti (569232) | about 5 months ago | (#46823075)

And there are no digitized images or videos that include faces anywhere among the data they haul in?

NSA hauls in no more suspects than the NRO or NOAA, which is 0. They just listen and report.

Re:money (1)

fuzzyfuzzyfungus (1223518) | about 5 months ago | (#46823715)

Listening is 'hauling in data'(unless you are purely passive about it, which they aren't.) I didn't mention humans at all.

Re:money (1)

i kan reed (749298) | about 5 months ago | (#46822507)

I don't think conspiracy theories are needed to allege feature creep, though.

Shit (3, Funny)

GameboyRMH (1153867) | about 5 months ago | (#46822221)

Of all the singularity-ish technological advances I could be alive for, it had to be this one :-(

let's see... (2, Funny)

martiniturbide (1203660) | about 5 months ago | (#46822233)

Let's see if it recognize me with my Groucho Marx glasses. Can authorities relate me as being a Marxist for that?

Re:let's see... (4, Funny)

GameboyRMH (1153867) | about 5 months ago | (#46822329)

"Are you, or are you not a member of the Marxist Party!?"

"Please restate the question, I'll only answer one at a time!" *wiggles eyebrows*

So... (0)

Anonymous Coward | about 5 months ago | (#46822263)

the position of both eyes, the nose and the two corners of the mouth

If you can't see the eyes the nose and the two corners of the mouth it fails?
Wake me up when it can identify a criminal in a joker mask.

Re:So... (1)

rossdee (243626) | about 5 months ago | (#46822327)

"Wake me up when it can identify a criminal in a joker mask."

You don't need a computer for that, everyone knows the Joker is a criminal.

Re:So... (0)

Anonymous Coward | about 5 months ago | (#46823511)

Not always, sometime he, Batman, and other DC characters play tennis, golf, or kart race.

Re:So... (1)

davester666 (731373) | about 5 months ago | (#46825877)

Obviously, you need to use a George Bush mask. He's a known criminal and they know where to find him...

Re:So... (0)

Anonymous Coward | about 5 months ago | (#46822339)

It will recognize the person in a joker mask as Jack Nicholson.

Code (-1, Troll)

Anonymous Coward | about 5 months ago | (#46822323)

Where is the code that does this analysis? If I don't see the code I am extremely doubtful. Especially from a Chinese University.

once our new mirrors are delivered we'll get it? (-1)

Anonymous Coward | about 5 months ago | (#46822333)

captive audience included with every bathroom based audition.. if i present well to myself how can i fail you? learning how to 'look' at ourselves differently & everything else changes too? what a gig.. little miss dna cannot be wrong.. meeting our own needs is unavoidable? good sports with good spirits wake up call

Er.... (0)

Anonymous Coward | about 5 months ago | (#46822347)

1984 was meant to be an AWFUL WARNING.

Not a manual for future technology development...

Re:Er.... (1)

Time_Ngler (564671) | about 5 months ago | (#46822911)

Then DO SOMETHING!

Re: Re: Err... (0)

Anonymous Coward | about 5 months ago | (#46825815)

I DID.

I went and joined the intelligence community. Seemed the most sensible way to survive...

There is only one word for this: (0)

Anonymous Coward | about 5 months ago | (#46822357)

scary.

t1000 (0)

Anonymous Coward | about 5 months ago | (#46822485)

One step closer to building terminators

I really doubt it does... (2)

Karmashock (2415832) | about 5 months ago | (#46822491)

I'm sure a computer is better at recognizing people that people don't know. That is taking in a database of 10,000 photos and matching them against a population set of 20,000,000. A human being really can't do that. However, face recognition of known faces? I really doubt the computer can beat us.

Now, you might say the computer has an advantage in that it can look at facial statistics and match faces even after plastic surgery but that's just because those alteration technics are designed to work against human beings.

If the primary threat is a computer recognizing you then you can alter your appearance in other ways to trick the machine. And those are technics that are unlikely to be as effective against a person.

Furthermore, using such methods is unlikely to be as taboo in human company as doing something that fools humans.

Re:I really doubt it does... (4, Interesting)

BasilBrush (643681) | about 5 months ago | (#46822745)

For sure, disguises to baffle algorithms differ from disguises to baffle humans. Here's a web site about disguises to baffle facial recognition systems. Probably not something anyone would want to wear outside a fashion event or a political demo, but interesting anyway.
http://cvdazzle.com/ [cvdazzle.com]

Re:I really doubt it does... (2)

Karmashock (2415832) | about 5 months ago | (#46823363)

That's just one type... there are lots. You can wear glasses that break up the symmetry of your face. You can wear hats that confuse it. There are a lot of things.

mod doi3n (-1)

Anonymous Coward | about 5 months ago | (#46822577)

I ever did. It if desired, ]we

When I'm ready for the rest home... (1)

ThatsDrDangerToYou (3480047) | about 5 months ago | (#46822641)

As long as I can have my implanted device remind me that, yes, it's my daughter visiting today. Her husband's name is X and they're kids names are... Technology will totally save my ass someday.

Re:When I'm ready for the rest home... (2)

BasilBrush (643681) | about 5 months ago | (#46822769)

Exactly the same technology would be very useful for salesmen.

Re:When I'm ready for the rest home... (0)

Anonymous Coward | about 5 months ago | (#46822997)

And also very useful against a salesman.

Salesman HUD:
BasilBrush: cynic, distrustful of salesmen. Use humble approach C7

BasilBrush HUD:
Salesman detected: will probably try to appeal to your ego. Suggest applying tissue damage to left leg. Knives are aisle 3, viable bludgeons are aisle 8.

Re:When I'm ready for the rest home... (1)

ThatsDrDangerToYou (3480047) | about 5 months ago | (#46823143)

And also very useful against a salesman.

Salesman HUD: BasilBrush: cynic, distrustful of salesmen. Use humble approach C7

BasilBrush HUD: Salesman detected: will probably try to appeal to your ego. Suggest applying tissue damage to left leg. Knives are aisle 3, viable bludgeons are aisle 8.

Clearly, the fist of death is the only solution here.

I was hoping for a prosopagnosia self-test. (2)

jeffb (2.718) (1189693) | about 5 months ago | (#46822667)

I was hoping the link would include an actual self-administered test, with your score reported at the end.

I looked through the "same" pairs a bit. It confirmed that I'm terrible with unfamiliar faces -- on perhaps a third to a half of the pairs, I would've had no idea that both represented the same person. For faces that I've seen hundreds or thousands of times -- Condi Rice, Rumsfeld, Bono, Jimmy Carter -- I guess I do as well as anyone.

For the "different" pairs, I only looked through the first dozen or so, and they all seemed obviously different to me. Also not surprising; I get lots of false negatives (failing to recognize a face), but few false positives (thinking I recognized someone when I really didn't).

I wonder how my life would've turned out differently if I'd grown up with a prosthesis to help me recognize faces. I wonder how much difference it would make if I got one today. I feel like I've come up with reasonable coping strategies, but I wonder...

Ok great... (1)

jellomizer (103300) | about 5 months ago | (#46822691)

Now where can we get it... Or purchase it.
Is it patented, is it Open Source.
Where is the link to the actual algorithm.

Re:Ok great... (1)

BasilBrush (643681) | about 5 months ago | (#46822837)

Now where can we get it... Or purchase it.
Is it patented, is it Open Source.
Where is the link to the actual algorithm.

The second link is to the article. It's available as a PDF. Only use it for good.

Re:Ok great... (0)

Anonymous Coward | about 5 months ago | (#46825383)

Now where can we get it... Or purchase it.
Is it patented, is it Open Source.
Where is the link to the actual algorithm.

No need to purchase it. You can find the algorithm here:
http://arxiv.org/pdf/1404.3840... [arxiv.org]

Idiots (1)

koan (80826) | about 5 months ago | (#46822927)

Complete and utter oppression of humans moves one step closer, soon no escape ever.

Not that hard to beat me (1)

BlackPignouf (1017012) | about 5 months ago | (#46823053)

It's really not that hard to beat me : I say "Hello, my name is ..." twice to almost everyone at social events.
Bogosorting pictures would do a better job than me.

Epic Fail! (it's really an urban legend) (1)

davecb (6526) | about 5 months ago | (#46823099)

If you have a 97.54% chance of a "perfect" match on all those criteria, remember the birthday paradox. If you compare 23 people against one another, you get (23*22)/2 = 253 comparisons. Multiply 253 * 0.9754 and you get 246.7 correct, 6 wrong.

Six failures out of a pool of 23 (hijackers + passengers) is insane, and is therefore one of the reasons that the German Federal Security Service rejected my employers' facial recognition system many moons ago. Until we get to 1.0, the number of false positives will be more than we can manually re-inspect in time for them to make their plane.

This deserves to be widely labelled as an urban legend, and ridiculed.

Re:Epic Fail! (it's really an urban legend) (0)

Anonymous Coward | about 5 months ago | (#46823797)

Inside any pool of people the result of every comparison will be negative, mate.

Re:Epic Fail! (it's really an urban legend) (1)

davecb (6526) | about 5 months ago | (#46825255)

Could you expand on this, please? On the face of it, you appear to be saying the probability of success will be 0.0 for any finite group...

Re:Epic Fail! (it's really an urban legend) (1)

viperidaenz (2515578) | about 5 months ago | (#46827445)

If you're looking for terrorists, it much worse than what you say.

How many terrorists are there really?
For arguments sake I'm going to say 10,000 in USA, with 320M population.
To get the number of false positives on the general population down to the number of real terrorists, so only 50% of the people you stop are innocent, you need a 99.996875% accurate test.

With 98.52 as claimed by this test, you'd have 4.7M false positives and 148 terrorists get false negatives.

This could mean one of two things (1)

Minwee (522556) | about 5 months ago | (#46823131)

Either facial recognition software has gotten a whole lot better, or the legalization of marijuana has started having an effect on people.

I'm willing to believe both stories.

Fishy Article (1)

Anonymous Coward | about 5 months ago | (#46823279)

Such a good result would easily get published on a high impact journal or conference. If they are self-publishing on arxiv, there is something very fishy about the research. No proper review, no source code, and chinese authors (known for flooding journals with incorrect bullshit in hope of gettings something publushed by accident).

The ultimate test... (0)

Anonymous Coward | about 5 months ago | (#46823447)

... would it be able to recognize me?

Picture related: :-)

based on five image landmarks: the position of both eyes, the nose and the two corners of the mouth

Channel Surfing (1)

esperto (3521901) | about 5 months ago | (#46823453)

Now the next big challange for AI will be channel surfing. I doubt it will ever change channels as fast as men do and still know what is being shown on each channel.

Only Outperformes Humans in 1 of 5 Datasets (5, Informative)

HannethCom (585323) | about 5 months ago | (#46823801)

Both the title and description are sensationalism. If you read the paper, the algorithm works at 98.52 on 1 of 5 data sets. We could also accurately say the algorithm is only 89.33% accurate. The score from the most difficult database. Much worse than the normal person.

That being said, it is much better than other publicly available algorithms.

Worship Shiva to defeat Gaussian Face. (3, Interesting)

140Mandak262Jamuna (970587) | about 5 months ago | (#46823973)

One of the most prominent features of Lord Shiva, is His third eye. [google.com] Paint a third eye on the forehead to completely discombobulate Gaussian face.

The other stunning form of Lord Shiva is His half-female version [wordpress.com] . If you could manage this form, you would discombobulate not just Gaussian Face but also fellow humans too.

Extending the theory, painting random noses, lips, eyes and other features on cheeks, foreheads etc would defeat these automatic face recognition systems.

Sensationalist summary but good research (1)

KramberryKoncerto (2552046) | about 5 months ago | (#46824141)

It's a lot of rather common techniques put together, but it's a meaningful result. But the summary is rubbish: on the same benchmark for which this algorithm gave 98.52%, there was an older algorithm that gave 96.33%, and a graph showing small incremental improvements up till then; saying that older results "never came close" is a blatant lie.

The methodology seems legit, in the sense that it's consistent with those used for previously accepted results. On the other hand, I still have the same old concern on all machine-learning type research. I believe the model is useful, but getting experimental results like this is about tweaking details to fit the same test data, and as soon as the accuracy hits a new high you can publish. There are certainly ceilings that a model can't break, and the results certainly show that their model has a higher ceiling for a specific dataset than before, but it doesn't prove that the algorithm is universally better than previous ones, certainly not humans.

Margin of Error (1)

medv4380 (1604309) | about 5 months ago | (#46824943)

The machine margin is probably fairly exact, but measuring humans gets a bit fuzzy. What is the margin of error on that 97.53 percent for humans. If it's 3% then I'm going to doubt the claim.

A grateful nation (1)

ThatsNotPudding (1045640) | about 5 months ago | (#46825211)

of Killbots thank you.

"Science: we're all about coulda, not about shoulda." -- Patton Oswalt

Eh (0)

Anonymous Coward | about 5 months ago | (#46825265)

Casinos have been using advanced facial recognition for years now.

Easy to set up patsies. (1)

140Mandak262Jamuna (970587) | about 5 months ago | (#46826255)

Imagine a tech savvy guy planning to commit some serious crime. He/She could find a patsy, take a few photographs on the sly. And get to work at home subtly adding makeup to make the eye slightly apart, or contact lenses with off-center pupils, add lipstick to make the lips slightly longer etc. Keep altering it tell Gaussian Face declares that new face of the criminal and the photo of the patsy both hash to the same 150 pixel canonical image. Now, commit the crime in full view of the security camera, and the police will go after the patsy.

Re:Easy to set up patsies. (1)

viperidaenz (2515578) | about 5 months ago | (#46827387)

Except I doubt 98% accuracy is admissible in court.

Say there are 320,000,000 people in a country, like USA.
If 5% of those people are criminals, that's 16,000,000 criminals.

With 98% accuracy against the 304,000,000 innocent civilians, that's 6,080,000 false positives, nearly half of the number of real criminals. That's only 62% change of successfully identifying a real criminal, despite a "98% accurate" test.

Re:Easy to set up patsies. (1)

140Mandak262Jamuna (970587) | about 5 months ago | (#46834323)

The goal is not to get the patsy convicted in a court. Just to draw a red herring across the trail of the police.

Cool (1)

viperidaenz (2515578) | about 5 months ago | (#46827299)

So to fool the face recognition of the future I just have to change the distance between the corners of my mouth.... by smiling or frowning?
If I perform all my crimes with a cheeky grin, they'll never match it to my frowny looking drivers license photo.

Good enough (1)

volmtech (769154) | about 5 months ago | (#46829119)

It might not be as good as a human but quantity has a quality of it's own. If cameras can pick out likely suspects then a human can take a closer look from the same feed.

"Finally"? What, is this is a good thing? (0)

Anonymous Coward | about 5 months ago | (#46830429)

"The panda bear is finally extinct!"

And Superman? (1)

blootsvoets (1090643) | about 5 months ago | (#46831379)

Ah, so would it recognize that Clark Kent and Superman are actually the same person? Imagine the benefits for villains all over the world!

chinky's (0)

Anonymous Coward | about 5 months ago | (#46831399)

does it works for chinky's they all look the same to me

Check for New Comments
Slashdot Login

Need an Account?

Forgot your password?

Submission Text Formatting Tips

We support a small subset of HTML, namely these tags:

  • b
  • i
  • p
  • br
  • a
  • ol
  • ul
  • li
  • dl
  • dt
  • dd
  • em
  • strong
  • tt
  • blockquote
  • div
  • quote
  • ecode

"ecode" can be used for code snippets, for example:

<ecode>    while(1) { do_something(); } </ecode>