Slashdot Mirror
The Face Detector
Roland Piquepaille writes "Almost all human faces have common characteristics, such as two eyes and one mouth. Still, some people, affected by face blindness, cannot recognize one face from another one. So it's understandable that face recognition is a major challenge for computer vision systems. In "Facing facts in computer recognition,", the Pittsburgh Post-Gazette reports that a team from Carnegie Mellon University's Robotics Institute has developed a very accurate software to find faces within images. By analyzing only 768 pixels, the system can detect 93 percent of the faces in a set of images while falsely identifying four objects as faces. The Face Detector Demo is available online and you can submit an image for analysis and receive the results by e-mail. The technology will be used for security purposes, but also by digital photography companies who want to automatically reduce "red eye" effects. You'll find more details and references in this overview."
I can't tell people apart sometimes. I have a VERY hard time remembering someone's name, especially when they're Person #12 I Met On The Tour Of The Office out of The 37 People I Met On My First Day.
But am I afflicted with "Face Blindness"? NO! I have a shitty memory for faces, and that's it! I don't have some made-up malady that can be cured with thousands of dollars of useless medication advertised on TV!
The technology sounds cool. The culture of euphemism in the US just pisses me off, that's all.
"instead of the endless "let's use it in airports" crap,"
The "use it in airports 'crap'" happens because they're trying to get gov't funding to continue development. It's not crap to try to get ahead in this world by fulfilling a need. Don't like it? Point at Uncle Sam instead of the company trying to earn money.
*Note: I'm talking in general, not specifically about this company.
"Derp de derp."
"This is a prerequisite toward ubiquitous facial recognition. "
Also good for video compression. Imagine if the codecs detected where the faces in a video were and intentionally avoided compressing them to hell and gone.
"Derp de derp."
Actually, that raises an important point. This system can identify a face, but does it identify only human faces? After all, chimpansees have many of the same facial characteristics as some humans.
There's no way for this process to reliably determine something like race, either -- not that doing so is that desirable anyway. The characteristics that make up "an African American guy" are just not nearly as concrete as we think they are from day to day. I have a neighbor who thinks all the Somali people in my area are "Arabs." Her category is a little too broad. It seems to me like she's forcing certain expressions onto their faces, too, as part of her image of what "Arabs" are like.
People's minds love to categorize. Sometimes, a lot of the time, we force information into categories it doesn't quite fit. (Refer to: State Department intelligence from Iraqi exile organizations.) Even when the information is essentially noise, we try to sort it and sift it. As a result we persist in holding weird ideas: astrology, because the paper tells us something vague and we run the events we see past that filter.
We should expect our tools to share some of those biases and blind spots. As much as we might try to address that, we have the blind spots ourselves, so it's hard to know how to counter the problem.
"Fundamentalism" isn't about divine morality. It's about human authority.
Bruce Schneier points out the problem of false positives in his "Secrets and Lies", and you'll get the same argument in any freshman statistics class: If the target population (of identified terrorists in the country, or people with AIDS) is extremely small, the probability of a false positive is greater than the probability of a true positive. If this system is correct 93% of the time, it's wrong 7% of the time. How many terrorists are there in the average metropolitan airport? I'd say zero (on average) How many people would be incorrectly identified? 7%. An airport with, say, 2000 people in it would have 140 of them misidentified. Even if the average airport had 20 terrorists, the false positives or misidentified would greatly outnumber them. And since the system is wrong 7% of the time, one or 2 of those 20 terrorists would be misidentified as not being a terrorist.
give it to blind people so they can know who they're talking to
Have you never had someone say "Hi" after calling you on the phone.... and known who it is?
Still, some people, affected by face blindness, cannot recognize one face from another one. So it's understandable that face recognition is a major challenge for computer vision systems.
Face blindness just shows us that the specialized hardware we have for face recognition is so incredibly accurate that we rely on it completely and have no alternate methods of face recognition. When it's broken, other parts of our brain don't step in either because a) it's a hard task or b) they just don't have access to the relevant visual information. Face recognition could be totally simple, if this were our only measure.
There are no trails. There are no trees out here.
As you upload your images to your PC, it verifies you havent taken any pictures of 'banned' individuals... Such as movie stars...
If you have, it deletes the image, and e-mails the MPAA.
Though I'm joking, there is room for such abuse once you have to be authenticated even to view your own images.
---- Booth was a patriot ----