Slashdot Mirror

← Back to Stories (view on slashdot.org)

Microsoft's AI Judges Age From Snapshots, With Mixed Results

Posted by ryuzaki0 on from the are-humans-all-that-much-better? dept.

mikejuk writes: A Microsoft Research project that lets users upload photos and estimates their age and gender has attracted more attention than expected — not all of it complimentary. The How-Old.net site demonstrates of some of the capabilities of the Face API included in Microsoft's Project Oxford that was announced at Build. It may have been expected to be a source of amusement but instead it backfired when people started to upload their own photos and discovered just how wrong its estimates could be. It demonstrates not only that machine learning has a long way to go before it's good at estimating age, but also that machine learning may not be the most politically correct way to go about answering the question 'How Old Do I look'. It might be better to employ and algorithm that built in all the rules of how to make a polite answer to that request — such as always knock a decade off the age of anyone over 28. Perhaps this particular neural network needs to learn some social skills before pronouncing how old people look. However it is capable of telling some truths — a photo of Barak Obama in 2005 gives an estimated age of 46, close to his real age of 44, but just 9 years later in 2014 the age guessing robot places him at 65. It seems that Mr President aged 20 years in less than 10 years of office.

45 of 80 comments (clear)

  1. presidents age by Anonymous Coward · · Score: 1

    It has been shown that presidents age quickly in office.

    1. Re:presidents age by __aaltlg1547 · · Score: 1

      It has been shown that presidents age quickly in office.

      That's more perception than reality. Aside from the possibility of getting assassinated, it hasn't shortened their life expectancy noticeably.

    2. Re:presidents age by TWX · · Score: 3, Funny

      It's probably more accurate to say that Presidents look haggard and appear to be older than they are while in office. Probably something to do with the responsibility and the stress. Thing is in this Microsoft software, if it doesn't have a means to address the difference then it probably will err on the side of older.

      Someone did a faces of pornography shoot where they took headshots of pornographic actresses before and after their makeup was applied. I wonder how substantial the differences there would be with the subject at the same age.

      --
      Do not look into laser with remaining eye.
    3. Re: presidents age by Anonymous Coward · · Score: 1

      Breaking news: people in their 40's and 50's after 8 years look older Than they did 8 years before.

    4. Re:presidents age by NicBenjamin · · Score: 3, Interesting

      I think it's mostly hair.

      A Presidential candidate is trying to look like a young man, a virile man, a man with great energy to do great things. He's probably in his late 40s or early 50s. He may have some gray, but if it's a lot he's probably dying it the fuck away (you'll note most of the GOP field has no gray at all, despite the fact they're all in their late 40s at least, back in '08 the Dem field was just as non-gray). If he's a she of course she's dying the gray away (I love you Hillary, but we all know your hair today is not precisely the same shade it was back when you personally murdered Vince Foster and then arranged for the paperwork to say "suicide" back when I was 12).

      OTOH, as his term goes on a) non-gray hair becomes less socially acceptable (just imagine what Jon Stewart would have done with the tidbit that Dubya was dying), b) it's harder to claim that you actually aren't dying (Obama will leave office at 55, and he's on the young side), and c) he wants to look like the Alpha Male Silverback Gorilla who has Already Accomplished Great Things. So you remember a 50-year-old from the campaign with no gray, then he gets on HD TV at 57 with the normal level of gray for a 57-year-old, and you think :"Gee the past seven years have really aged our President," when in fact all that's happened is he's stopped using Just For Men.

    5. Re:presidents age by quantaman · · Score: 3, Interesting

      I think it's more an effect of the people we see pictures of most, celebrities, put a lot of work into appearing young, so we don't expect people in the public eye to age as quickly. Tom Cruise ages slowly because his career demands it, Barak Obama on the other hand probably looks more serious the older he looks, so there's less reason to make himself appear young.

      Even compare to Jon Stewart in 2008 vs now. There doesn't seem to be a huge difference, until you realize the gallon of makeup applied to Jon Stewart's face, it's hard to appeal to GenXers and Millenials looking like you're over 50.

      I'd actually be curious to see how this algorithm does with celebrity photos.

      --
      I stole this Sig
    6. Re:presidents age by Mikkeles · · Score: 1

      ...people we see pictures of most, celebrities, put a lot of work into appearing young...

      You mean like this?

      --
      Great minds think alike; fools seldom differ.
    7. Re:presidents age by hairyfeet · · Score: 1

      Good Lord boy, even I think Hillary is a DINO that is more of a warhawk than even Dubya and will make another Dubya level of shitty when it comes to being the POTUS but even Ken Starr ruled that Foster killed himself, and that was after both the Parks Dept AND the FBI came to the same conclusion, what more do you need?

      As for the POTUS looking old? Stress ages people, everybody knows that and the POTUS is a stressful job. Hell look at what old Jimmy Carter looked like when he was sworn in and compare him to the 1980 election, the man looked like he aged a decade in those 4 years.

      --
      ACs don't waste your time replying, your posts are never seen by me.
    8. Re:presidents age by NicBenjamin · · Score: 1

      What more do I need? Fellow Slashdotters who can detect a joke. That's what more I need.

      Nobody who thinks the Starr investigation was a whitewash is going to start that sentence with "I love Hillary but..."

  2. Well it's difficult by aaaaaaargh! · · Score: 5, Interesting

    The only guy I've ever met who could judge the age of people well was a man who made some money on the Stroget in Copenhagen more than ten years ago by betting with pedestrians that he could guess their age (he probably did it mostly for fun, though). But he didn't use the face only, he used all kinds of clues, especially he looked at people's necks. He was really good, but everyone else just sucks at it.

    No wonder A.I. can't do it when ordinary people can't do it either.

    1. Re:Well it's difficult by kmahan · · Score: 2

      Looking at people's hands can also be a good datapoint for age.

      --
      Invalid Checksum. Retrying.
    2. Re:Well it's difficult by CastrTroy · · Score: 1

      People are inherently bad at this. I get asked for identification all the time when purchasing alcohol and I'm 35. The drinking age is 19. They are supposed to ask anyone who looks under 25, but I think they are trying to change it to ask anybody under 30 because they realized how hard it is to judge age.

      --

      Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
  3. GIGO by OzPeter · · Score: 1

    If people can't tell your age when looking at you, how can you get a computer to do it accurately?

    A persons apparent age is going to depend on a whole bunch of things including environmental things. EG take a farmer who spends all of his time outdoors and someone who partakes in a regime of moisturizers, skin peeling and other assorted beauty activities and lives in their mom's basement. Even if they are the same biological age, they won't look the same apparent age.

    --
    I am Slashdot. Are you Slashdot as well?
    1. Re: GIGO by Anonymous Coward · · Score: 1

      You seem to assume that computers can't be better than humans? I can say that this software is already 100 times better than me. Only flaw that I saw was that it can't detect faces from cartoon characters. But that is obviously because it was not trained to do so.

    2. Re: GIGO by OzPeter · · Score: 2

      You seem to assume that computers can't be better than humans?

      Let me simplify my comment for you: A computer cannot control for factors that are not encoded in an image. But those factors can effect apparent age.

      --
      I am Slashdot. Are you Slashdot as well?
    3. Re: GIGO by itzly · · Score: 2

      On the other hand, a computer can be trained to pick up clues in the image that most people wouldn't notice.

    4. Re: GIGO by OzPeter · · Score: 2

      On the other hand, a computer can be trained to pick up clues in the image that most people wouldn't notice.

      In which case it should have correctly estimated Obama's age in the second picture.

      Or is this a "true scotsman" issue in that "a well trained system would have correctly estimated obamas age in the second pic" ??

      --
      I am Slashdot. Are you Slashdot as well?
    5. Re: GIGO by Rei · · Score: 2

      I agree. Duh, the program is obviously not perfect and screws up sometimes. But I'm amazed by how good it actually is. Even being able to just ballpark it some of the time would be impressive, but the fact that it gets pretty reasonably close most of the time, I find that incredibly impressive.

      Someone on my Facebook feed was complaining about how in a washed-out picture of three children the picture guessed only two of them right, but saw one (a young boy) as an adult woman. My response was to crop out just the washed out face, take it out of context, and point out, if you saw this face, not understanding anything about the context, could you guess it? I certainly couldn't have. But that's exactly what the software has to do.

      I took a number of pictures of myself in different angles, making different faces, etc, and its range on age guesses was only 3 years. My brother-in-law managed to get a 20-year difference in guesses by making faces, but I couldn't manage it, and neither could most people I know who tried. Again, computationally, it's very impressive.

      --
      Sigur RÃs: I didn't know that Heaven had a rock band.
    6. Re: GIGO by angel'o'sphere · · Score: 1

      Only if the programmers know about those clues, which they likely don't. As "ordinary humans" or "doctors" don't know them either.

      I'm 48 ... most people guess me 35. Now teach them to guess me close to my age ... no idea how you would do that, even less how you teach a computer to do that.

      --
      Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.
    7. Re: GIGO by Dog-Cow · · Score: 1

      That's a ridiculous argument. Context matters. If you, viewing someone in person, have more context, it's not unreasonable to expect you to do better. The question you should be asking is: can you do better than the computer with just a photograph and no other knowledge?

      I suspect the answer is yes, but for how long will that be true?

    8. Re: GIGO by unrtst · · Score: 1

      What's with the surge of "no true scotsman" comments?

      Anyway... that doesn't apply here because you are mixing up tenses. Itzly's "can be trained" versus OzPeter's "should have". Also, rather than "a well trained system would have correctly...", it is "a well trained system will correctly ...", and that's assuming "trained" doesn't only mean "fed an accurate and large corpus" but also includes algorithmic tweaks/training.

      Either way, I think the quality of the photos would likely play a HUGE role. Garbage in, garbage out, and all that. If it can see the pores in your skin, it can do loads more detecting than the naked eye, but when they're using tiny little multi-portrait photos with a total file size capped at 3mb, that's quite a restriction on what it has to work with... and I'm shocked how well it does for the most part.

  4. Wedding Pictures by Cassini2 · · Score: 1

    I ran some wedding pictures through the site. By the end of the evening, the bride and groom were 15 years younger! It was a good wedding.

    1. Re:Wedding Pictures by Sique · · Score: 1

      I used a pair of pictures of me as a child, which were taken when I was about 13. They look nearly similar (two shots out of a sequence done by a professional photographer). The AI estimated my age at 26 for the first and 42 for the second one.

      --
      .sig: Sique *sigh*
  5. Re:This is ridiculous by epyT-R · · Score: 1

    or better yet, they should work to improve what it is they're self conscious about.

  6. I find it hilarious; Big Brother aspects less so by sideslash · · Score: 2

    C'mon, it's a fun tech project, and people should stop whining about it.

    What's not funny is where we are going with this kind of technology. Always on facial recognition and people tracking is already happening some places in the world. Casinos and airports were probably the leaders in the field, but soon it will be everywhere. I despise the thought of a government database tracking people everywhere, not to mention law enforcement reliance on face recognition that (as evidenced here) may be less than perfect. "The computer says you did it, so you're guilty."

  7. an example by wbr1 · · Score: 2

    http://imgur.com/gallery/Q1t0T...

    --
    Silence is a state of mime.
  8. Re:Hilariously bad. by quantaman · · Score: 1

    A relatively normal looking friend of mine in his mid-twenties tried it out.

    He came out as a woman in her 60's.

    --
    I stole this Sig
  9. unfair criticism by u19925 · · Score: 3, Insightful

    The criticism of website is unfair. Some people in some photos indeed look younger. Instead of criticizing AI algorithms, I would suggest take a test by using a random photo where you don't the true ages beforehand but have a way to find it out. Compare you guess with the website and see who does better. I tested several photos from my collection by asking my colleagues to guess the ages of my extended family members and the website. On an average the website was actually better than my colleagues.

    1. Re:unfair criticism by TechNeilogy · · Score: 1

      The criticism reminds me of the Gary Larsen cartoon where the guy is scolding his dog for doing a poor job mowing the yard.

      --
      "The wisdom of the Patriarchs was that they *knew* they were fools." --Master Foo
  10. Re:I find it hilarious; Big Brother aspects less s by Headw1nd · · Score: 1

    So we take something with the same success rate, and give it a veneer of science and technology?

  11. Square Enix by ArcadeMan · · Score: 1

    It seems Square Enix did an excellent job with the Miqo'te race of Final Fantasy XIV:

    http://moetoy.com/wp-content/u...
    http://geeksprinkles.com/wp-co...
    http://24.media.tumblr.com/849...

    Anyone dare to try that AI with the other races, especially Lalafell? :p

    --
    Get free satoshi (Bitcoin) and Dogecoins
  12. Hair color and lighting by Pfil2 · · Score: 1

    I tried several photos. I got as low as 23 and as high as 46 with photos from within the past 3 years (I'm in my mid-30's). In one photo I'm standing next to my mother who is in her late 60's in the photo. In that photo it says I'm 46 and she's 40. I believe the difference is that my hair is starting to go gray but my mother still colors her hair. In the picture where it says I'm 23 my hair is cut short and isn't noticeably going gray. However, I used the exact same picture and adjusted the lighting and it changed my age from 23 to 27.

    Curiously, it seems to have a problem with infants. I have a 22 month nephew that depending on lighting says is either a 5 or 16 year old girl. I also tried a picture of my nephew being held by his mother at only a few months old and it guessed he was 5 years old. The algorithm is definitely using just the face as if it used other contextual clues it'd be very obvious a 5 year old is much larger than a 3 month old or that a 16 years old is much taller than a 5 year old.

    Interesting concept but it appears hair color and lighting play large roles (simply based on my own manipulation of identical pictures).

  13. I got burned once... by dargaud · · Score: 2

    Before I met my GF's mom, she told me that her mom looked much younger than she was, without telling me her age (or I forgot). So when I met her, she proudly asked me how old I thought she was. As a good programmer, I compensated knowing the way the bias went, and gave her a good decade more than she actually was. She didn't appreciate. Yeah, social skills, blah, blah, blah...

    --
    Non-Linux Penguins ?
    1. Re:I got burned once... by Anonymous Coward · · Score: 1

      the birth mother of your girlfriend is always 18 years older than your girlfriend (less if you know for a fact she gave birth very young) even if she was born to a 45 year old mom; and always looks 10 years older or less (so they can look like 'sisters'. for some reason moms love thinking that).

  14. Same with people by wonkey_monkey · · Score: 1

    Microsoft's AI Judges Age From Snapshots, With Mixed Results

    The same mixed results you'd probably get with people doing it.

    Just look at those two photos of Obama - he really does look older in one than the other. If you didn't know who he was, or if you found similar looking images of two distinct people, those age guesses would not seem wildly inaccurate.

    --
    systemd is Roko's Basilisk.
  15. Other websites it is better not to build by SuperKendall · · Score: 2, Funny

    DoILookFatInThisDress.com

    AmIPregantOrJustFat.com

    ShouldIWearSpadex.com (well you can build that one but only if it always returns NO for all input)

    HighlightLinesOnMyFace.com

    UsePhotosFromThisFacebookFeedToPhotoshopSomeoneCheatingOnSO.com

    --
    "There is more worth loving than we have strength to love." - Brian Jay Stanley
  16. Disappointing by PopeRatzo · · Score: 1

    I thought this sounded like fun so I uploaded my photo to the Microsoft age-guessing app.

    All it said was, "Shall we notify next of kin?"

    I think it must be broken.

    --
    You are welcome on my lawn.
  17. Age as a statistical concept by aNonnyMouseCowered · · Score: 1

    I read somewhere that not only do people age differently but different parts of the same person age differently. But just what is age, if not meausred against some arbitrary clock? Just think of Einstein's thought experiment about relativity, the astronaut twin who ages a few years vs the earh-bound twin who ages thousands of years (assuming he has someone managed to find a life-support system that makes him surive to that age).

    But if we assume the mundane notion of age being something we celebrate each year or put in a person's obituary or Wikipedia page then clearly no visual system or medical diagnostic can accurately calibrate a person's calendar age. The lungs of a forty-year old smoker who never exercises may well look and feel older than the lungs of an eighty-year-old who jogs regularly and plays the saxophone in the local orchestra. So yes maybe Obama does look 70 years old, visually at least, when compared to a sample of photos of people born on this planet 70 years ago.

  18. Term of Use by PixetaledPikachu · · Score: 3, Informative

    This is from the website's TOS, which happens to be the TOS for Azure as well:

    "Microsoft does not claim ownership of any materials you provide to Microsoft (including feedback and suggestions) or post, upload, input, or submit to any Website Services for review by the general public, or by the members of any public or private community (collectively "Submissions"). However, by posting, uploading, inputting, providing, or submitting your Submission, you are granting Microsoft, its affiliated companies, and necessary sublicensees permission to use your Submission in connection with the operation of their Internet businesses (including, without limitation, all Microsoft services), including, without limitation, the license rights to: copy, distribute, transmit, publicly display, publicly perform, reproduce, edit, translate, and reformat your Submission; to publish your name in connection with your Submission; and to sublicense such rights to any supplier of the Website Services."

    So, how are you, future microsoft's advertisement stars?

    1. Re:Term of Use by Pfil2 · · Score: 1

      On the site under the "Use your own photo" button it very explicitly says "P.S. We don't keep the photo".

      Then on the blog it says:

      "Updated 5/2/2015

      We've had some questions so we updated this post to be more clear. To answer the top one: No we don't store photos, we don't share them and we only use them to guess your age and gender. The photos are discarded from memory once we guess. While we use the terms of service very common in our industry, and similar to most other online services, we have chosen not to store or use the photos in any way other than to temporarily process them to guess your age."

      So, if you become a Microsoft star you probably have a very good legal case. IANAL.

    2. Re:Term of Use by radarskiy · · Score: 1

      The thing is you really do need those rights grated to you in order to run a site like that, since the service does all of those things. Otherwise someone could upload their photo and then sue you for copyright violations even though they were the one who gave you the photo.

  19. Re:Hilariously bad. by Dog-Cow · · Score: 1

    I imagine the porn stars are trying to look underage. Is it the program's fault it doesn't have the larger context?

  20. It has problems with age but by Wescotte · · Score: 1

    it seems to be incredibly good at picking the right gender which I have to admit is pretty impressive.

  21. If corporations are people by Tablizer · · Score: 1

    I fed it an MS brochure, and it said the company acts 80 years old

    --
    Table-ized A.I.
  22. Correcting the guess by Mike+Van+Pelt · · Score: 1

    Interesting -- someone in the office was running everyone's picture through this page, and if you averaged their guess of my age (52) with Microsoft's guess (68. Gee, thanks ever so much, Microsoft) you get my actual age. (60)

    I'm surprised there's no "Wrong, the true age is..." input, to help them correct their model.