Forgot your password?
typodupeerror
AI News Technology

Face Recognition Maps History Via Art 36

Posted by Soulskill
from the computers-confused-by-lack-of-eyebrows dept.
mikejuk writes "Face recognition techniques usually come with a certain amount of controversy. A new application, however, is unlikely to trigger any privacy concerns — because all of the subjects are long dead. 'FACES: Faces, Art, and Computerized Evaluation Systems' will attempt to apply face recognition software to portraits. Three University of California, Riverside researchers have just received funding to try and piece together the who's who in history. 'Almost every portrait painted before the 19th century was of a person of some importance. As families fell on hard times, many of these portraits were sold and the identities of these subjects were lost. The question we hope to answer is, can we restore these identities?' If the algorithm can be fine tuned we can look forward to the digitized collections of museums and art galleries around the world suddenly yielding a who-knew-who social network graph that could put more science, and computer science at that, into history."
This discussion has been archived. No new comments can be posted.

Face Recognition Maps History Via Art

Comments Filter:
  • by busyqth (2566075) on Saturday April 28, 2012 @05:23PM (#39833685)
    Think of the possibilities...
    If I could know for sure that the eldest daughter of the 1st Earl of Huntshire was a good friend of the young wife of the wealthy merchant heir James Strickthorpe, well... it would completely change my life.
  • Facebook (Score:3, Insightful)

    by programmerar (915654) on Saturday April 28, 2012 @05:26PM (#39833699) Journal

    Waiting for Facebook to auto enroll them, give them a timeline and a social graph.

  • I wonder if it will allow to poke some prussian chicks.
    • by busyqth (2566075)
      Not being sure what you were implying, I image searched for "prussian chicks".
      Now you and I may have different tastes, and I can't say that the uniforms were all that bad, but the moustaches and rifles are kind of off-putting in my opinion.
  • I'm dead and I want privacy too, you insensitive clod!

    • by JWSmythe (446288)

      There's a real case there. Imagine a vampire, who has changed his identity every 20 to 30 years for centuries. He'd fake his own death, and move on with enough riches to start over.

      Now they'll have a lineage of the names he used, where he lived, and quite likely be able to identify trends in his feeding patterns. The truth will be known, and there will be nowhere for him to hide. The lineage of his crimes will haunt him forever.

      I guess the important part of

  • a good portrait is more about capturing a persons persona than their likeness. A portrait that sells is all about gilding the lilly.
  • I'm sure to have relatives amongst that lot - and given how much of my life has been affected by stuff my ancestors did, I'm pretty confident that this will help.

    And I have an archive of photos.

    Many of our ancestors did amazing things. Sometimes wonderful, sometimes terrible - but definitely amazing.
  • by khipu (2511498) on Saturday April 28, 2012 @07:22PM (#39834127)

    That kind of project is a sham. Face recognition software works by precise geometric measurements and by identifying unique and precise skin patterns. Neither of those are present in paintings. Paintings vary a lot more and still require human abilities to interpret facial characteristics.

    • Would you mind expanding on what humans use for face recognition that isn't yet present in software algorithms? I'm sure you know what you're talking about but it seems counterintuitive to me when I perceive most painted portraits to reproduce faces just as faithfully as a candid, poorly-lit Facebook photo, the latter of which is easily recognised by software.
      • by khipu (2511498)

        Face recognition algorithms (at least the ones that work well enough for practical use) recognize faces based on exact shape and appearance. Humans often use characteristics and qualitative deviations from "normal" faces. That's why humans have no problems recognizing caricatures like these: http://tinyurl.com/d8pq9f6 [tinyurl.com] Computers can't do that yet. Paintings tend to be more like caricatures, not photos.

        And face recognition in Facebook usually only has a few dozen people to choose from, with a high proba

        • by chrismcb (983081)

          Paintings tend to be more like caricatures, not photos

          Yes you are right, I could barely tell the difference between those cartoons and the Mona Lisa, or any other portrait.

        • by chrismcb (983081)

          And face recognition in Facebook usually only has a few dozen people to choose from, with a high probability that the same faces occur again. In art, you have tens of thousands of faces and a low probability that any one occurs multiple times.

          Are you trying to say that there are only a few dozen users on facebook? My picasa album has close to 200 people I've identified, and several thousand other people that I haven't bothered to give names to.
          It seems to me that the real issue is when there is only one or two paintings of one person. But it still seems like a fascinating research topic.

    • by chrismcb (983081)

      That kind of project is a sham. Face recognition software works by precise geometric measurements and by identifying unique and precise skin patterns. Neither of those are present in paintings. Paintings vary a lot more and still require human abilities to interpret facial characteristics.

      That must explain why picasa keeps recognizing artwork and sculptures in my pictures.

      • by khipu (2511498)

        You're confusing face detection and face recognition.

        • by chrismcb (983081)
          Picasa does a pretty good job with grouping similar faces together as well. It is perfect, but I assume a project like this would have some better algorithms.
          • by khipu (2511498)

            Why would you assume that "a project like this would have better algorithms"? Do you think that art historians have some storehouse of algorithms that are just unknown to Google's stable of top engineers and scientists?

  • by Anonymous Coward

    ...but it may actually identify the artist. It is a well-known phenomenon that portraits often have facial characteristics more closely associated to the *artists* face, not the subject.

    http://www.guardian.co.uk/artanddesign/2008/sep/07/art.heritage

  • I wonder if the technique can be applied to, ummm, other bodily parts?

  • This research is really IN YOUR FACE!

    Terribly sorry. Have a nice day!

You will be successful in your work.

Working...