Slashdot Mirror
Face Recognition - Real or Science Fiction?
An anonymous reader writes "Facial recognition software has been touted as one of the technologies that will change our future, particularly in law enforcement. How close are we to being recognized by a computer anywhere we go, as portrayed in movies like Minority Report? According to the industry's recent Public Relations releases, these products are closer than we think.
The reality though, is that current products work only when utilizing a small comparative sample, and any attempts for an individual to disguise themselves typically throw off the results. To see how far this technology needs to go before becoming mainstream, one site utilized Government-tested face recognition software, available freely through MyHeritage.com, to compare hundreds of famous people, animals, and cartoons to a database of 2,000 celebrities. Some of the results showed promise for the technology, but most were just funny — for example, who would mistake Barbara Streisand for Shrek, or Lance Bass of N'Sync for a Teletubby?"
"who would mistake Barbara Streisand for Shrek, or Lance Bass of N'Sync for a Teletubby?"
I think it's more a question of 'how many beers' than of 'who.'
After working in computer vision for 5 years I've realized that most problems aren't hard - they are not well defined. Mathematically face recognition is not a problem that can be stated.
Many other problems in CV are like this - edge detection, segmentation, etc. But people write hacks that work in restricted conditions and say they've solved.
And look, you could always just put on those Groucho Marx glasses.
This is all well and good, but the minute I get falsely identfied as a criminal just for being in the bar district late at night in the wrong place/wrong time I won't be too happy. . .
disclaimer: I've been known to store numbers in my ass for which to dig out when quantities are required.
So I guess next time a teletubby or Shrek wanders through a mall, they're totally going to throw off the face-recognition software.
Is it just me, or does that seem like a stupid way to test the software? If you want to show that rudimentary disguise is an easy way to get around it, that's valid, but just messing with the sample of potential matches by throwing in cartoon characters destroys the validity of the "study".
-stormin
The Southern Baptist Convention has creationism. On Slashdot, we have porn.
I thought they used chips in the eyes of people in minority report, not face recognition.
I've tried out the software and it was fun for some laughs. I'm not sure how it works exactly but I can tell that the angle of the face makes a difference. When I put one picture of myself in where I'm looking ever so slightly to the right, I'm matched with celebrities photos looking in that direction. When I put in a similar photo facing the other direction, I get a different set of celebrities looking in the other direction. There's a few overlaps and those are the ones I think I look the most like (although it's a stretch to say I have anything that could pass as a celebrity look).
Not to nitpick excessively, but you could easily substitute portions of this article with terms like (and relating to) “Internet”, “personal computer”, “telephone”, “car”, and others. Asking ourselves if a technology is “real or science fiction” when it already exists (albiet in a primitive form) is silly. Of course it exists; the question itself cites examples. Perhaps the meaningful questions might be along the lines of: “what are the challenges associated with making it accurate?” or “what impact will facial recognition have on society?”
Why bother.
I'm wondering about the legality of all this, especially in a criminal justice system. My DNA, for example, can't be used in court as evidence unless certain hoops have been jumped through; the prosecutor needs a reason to obtain a DNA sample and then procedures must be followed.
I wonder if the same systems will apply to a computer analysed image of my face; will there be a criterea for when this image is admissable in court? Will I have rights concerning my image? Or are we just going towards a 1984 style system. Interesting because this hasn't been the result of DNA admissions to court, despite the seemingly more robust nature of this evidence.
For example, who would mistake Barbara Streisand for Shrek, or Lance Bass of N'Sync for a Teletubby?
So, i see it's working correctly!
I believe Minority Report used retina scans, but that nit aside facial recognition works to a degree and will only get better. Security cams will eventually upgrade to HDTV resolutions, perhaps augmented with very high resolution stills when a potential match is made. This will all take more processing power, but all mighty god Moore will eventually gives us this day our daily CPU load.
About false positives. So what? Eyewitnesses make mistakes also. Eventually, perhaps very soon, machines will surpass humans in this arena just as they have in others. Can anyone here on Slashdot defeat Deep Blue at Chess?
As to the legality or ethics, what can be done will be done, at least in public areas. If it would be legal for a human to do (they haven't outlawed humans scanning for suspects in public areas) then it will be legal for machines to do despite the unease many will feel knowing they are constantly being watched.
Letter To Iran
But don't we almost always get a computer to solve a problem that's not strictly a mathematical one using "hacks that only work in restricted conditions"?
Our spell-checkers in our word processors don't actually know anything about the rules of a language, phonics, etc. They just do lookups from a dictionary. If a word's not listed, it has no idea if it's spelled properly or not -- even if the misspelling is one that's simply not a possible correct sequence of letters for the language. Most don't even realize if a word is misspelled in the context of the sentence, as long as it matches a correct spelling in the word list.
Until we figure out how the human brain recognizes faces as individuals, we can't expect anything *but* a clever hack for a computer to do the same. And truthfully, I suspect the human brain takes many things into account to do a "recognition" on a person. How often do you see somebody in the store that you're pretty sure you know from a previous job, school, etc. but you're not quite sure? I've had this happen a few times, and to make a better determination, I had to take other factors into account, like the sound of their voice if I heard them speak, the way they walked, or maybe an expression that came across their face. Humans "key in" on specific things that help them remember a person. And depending on which "features" they chose, they may or may not be effective. (Say you remember a gal really well because of her long, flowing hair? If she cuts it real short, there's a good chance you won't recognize her at all anymore if she walks by you.)
The problem is the inputs. Do you inputs sets of geometry (eyes are X" apart, at an angle of 0.53 degrees, chin is .5" below lips, blah blah blah), the raw image, or something else? If you use the raw image, you'd need a system in the front end scale/rotate the images to be in about the same place otherwise you probably have no chance (unless you want your neural net to do that TOO, which would make training harder and take longer).
Even if you use geometry (we have a vague understanding of what makes people look similar or beautiful) you'll still run into problem. You have problems of perspective (not all pictures are taken straight on).
Garbage in, garbage out. The best solution is to provide tons of information and let the neural net sort out what matters and what doesn't (they are quite good at that) but that will require more training which means more time.
So in the end you may build a good system. But to use it you must provide it with geometry of a face that someone picks out after fixing the perspective on a photo. Or it works much like our brains and accounts for all that, but it will take you 6 years of non-stop training alone.
And what is a success? Two people who look similar? A perfect match? What if your software rates a picture of a celebrity impersonator (looking like the celebrity) over a picture of that celebrity looking different (movie role, disheveled mugshot, etc)? Is that a success?
And how do you rate the people for the training input? Sure a neural net can figure out the way to something where we know the end, but what about when we don't quite know the end?
It probably took evolution a VERY long time to get good at recognizing individuals. And even then, we are not that great (mistaken identity, all cocker spaniels look alike until you spend more time with them, etc).
It's a neat problem, but it is seriously tough even with the "voodoo magic" that a neural net would provide over trying to come up with a straight formula.
Comment forecast: Bits of genius surrounded by a sea of mediocrity.
Every morning I wake up to look into the mirror and it's a different face that I don't recognized. Maybe I need to upgrade my mirror?
If I was training to match V1-4, I'd have the input come from two "eyes" with inputs similar to what our eyes actually provide to our brain. We know quite a bit about visual cortex, but there's a lot we don't know. Initially, I'd train it using a batch of photographs for a single person (we'll call her "Momma") and then I'd train with a few others (where a match is a match only if it's the same person). From there, I'd create histograms of parameter settings that seem to do an adequate job on this small set, and then use this reduced parameter space to create populations that are evaluated after training on millions of photographs. (The photographs can be placed in front of the eyes - once for each photograph, mind you, and not for each "individual" being tested - just like we can recognize photos and not just people.)
I could imagine narrowing the parameter space down to 100 or so unknown parameters, and each training session might take several hours. Given enough resources (e.g., the Pittsburgh Supercomputer Center), I'd run population sizes of 500 or so (in parallel), so that you could possibly go through 4-5 generations per day. In a month, you might have some pretty good individuals. Of course, my research area is the hippocampus and not the visual cortex, so it might take significantly more than 100 parameters to even begin to set this up.
Now, someone else pointed out that such computers would not have the biases that we humans have, but that's not necessarily true. If you train the computer using an input set of 950,000 "white" people and 50,000 "black" people, it would tend to make the mistake of thinking that "black" people look a lot like each other. (Studies done with speech recognition have shown that neural networks trained on Japanese have a much harder time telling "l" from "r" than those trained on English.)
Ben Hocking
Need a professional organizer?
"How close are we to being recognized by a computer anywhere we go, as portrayed in movies like Minority Report?"
Now I could be wrong but I am pretty sure Minority report was portraying retinal scanning not facial recognition
I'm afraid I'm going to call shennanigans on some of this. I've been doing Vision work for about 5 years now with a hefty does of image and signal processing in the mix(Working as gradstudent in the field right now in fact). Edge detection is well defined. The canny and shah-istan(think that's the name) are about as close to a mathematical optimal edge detector as one can get. There is in fact a well developed body of theory regarding differentiation of Signals. The problem doesn't lie in the mathematical models involved. It lies in how many people want to use those models. Edge detection suffers from spurious edges or edge flakes which are a symptom of noise in the signal at differention(ie differentiation enhances noise, integration smooths it). Segmentation can also be well defined you just have to be clear on what it is you're segmenting. Are you working in a color space, texture, motion? That matters. However you can get some very good results in these fields. See GPCA techniques for some examples of doing it. Or even modified PCA + EM or PCA+ Kmeans(clustering theory). Again very well defined. Mathematically there are several models for face recognition. One can examine the ideas of eigen faces(not my personal favorite but it's there), kernel based SSD type approaches to find key points, partial face detection followed by recognition over a sequence of images used to reconstruct the face, and more. The problem isn't the math. It's that when you project a model you are essentially destroying an entire degree of freedom which is a huge deal. Further just as you can match a partial finger print or a partial ear print you can match partial facechunks. The problem with makeup or facial hair comes when one relies on global matching techniques or uses only 2d information to do the matching. Now I'll be a first to say that alot of computer vision is a solution in search of a problem or that people do use a number of cheap hacks and dirty tricks to get things working but saying it's not mathematical is a lie. I can turn around and see at least 3 books at a glance that detail the mathematics that are a part of vision and image processing. So please don't confuse peoples fuzzy use or lack of understanding of the math for there being no math. Note: Machines are also bad at a number of tasks humans are really good at but the same can be said that there are many tasks that humans are very bad at but the machines excel at. Absolute range detection is a good example. Humans are very bad at telling you the exact range to an object, even with some sort of scale of the scene reference. Computers on the other hand(while suffering from noise in the signal) are still able to achieve significant accuracy depending on the range. You can see tyzx for an example of a comany who makes highly accurate stereo rigs.(They were around as of 2 years ago at least and I assume they're still going strong) Cheers
I don't care what you say, all I need is my Wumpabet soup.