0
Why Face Scanning Will Never Be Reliable.
Source
Why Face Scanning Will Never Be Reliable
From Ted Twietmeyer
tedtw@frontiernet.net
This is a subject that has bugged me ever since that failed, feeble attempt was made to face scan the entire Super-Bowl audience as they entered the stadium. Only incorrect identifications were made that day. In science terms, it was a total disaster. This is technology wearing a diaper with a brown stain. But that doesn't mean the concept of face scanning has died a quiet death. As long as there will be endless funding to increase government's grip on the populace, many more attempts will be made to kick this dead horse. If people make mistakes in police lineups, what chance does a computer have of working correctly which completely lacks fundamental human attributes such as real intelligence and common sense?
Take the main features of a human face: One mouth, two cheeks, a chin, two eyes, eyebrows and a nose. How many possible ways can you re-arrange these features? Since that arrangement was engineered into fixed positions a long time ago, there is only one way to arrange them (birth defects aside.) There are currently some SIX BILLION people on this planet. Are there really SIX BILLION possible faces? That idea is laughable at best.Now let's give face scanning a fighting chance and scale things back a bit. In the United States there is a population of about 280 million. Are there 280 million possible faces? Common sense shouts NO! While it is true that the spacing and proportions of each facial feature can vary, variations must be significant enough to cause visible differences. This also limits the smallest possible practical variations between people. For example, a difference in spacing between the eyes of 1/8 of an inch will hardly be very noticeable to anyone. And, not all facial variations are relevant either. And of course, there are the effects of aging for which software can make projections into the future, but can never be 100% accurate. In fact, there is an old expression that we all have a double walking around. A recent television show "That's Gotta Hurt" which aired on April 11th 2007, documented the life a lady who recovering from a very bad accident. Her husband is her caregiver, and he is a dead ringer for the actor Kevin Spacey. He has the same face, build, height, hair style, hair color, voice, everything. But it isn't him (the husband's name was given on the air.) Most likely there isn't just one clone of each of us walking around but several of them.
Human faces are all based on Phi, otherwise known as the Golden number or the Divine Proportion. Phi has a non-repeating value of 1.61803398874989. The human ear shown above is also based on this. It is true there are small variations in the ear's size, shape and spacing, but the overall design is the same for everyone. Genetic programming at work.Even our teeth are based on this Phi:
With these facts in hand let's return to face scanning. What happens when a man grows a beard? Can a face scanning system truly simulate what he looks like shaved? Some believe it can, and crime dramas would have you believe that. But - are all men's beards trimmed to conform to the shape of their face? There are numerous ways to grow and trim a beard. A man with a big, bushy overgrown beard can completely covers his chin and the sides of his face. What if a person's image in a database has a different beard style than what they have now? Will any face recognition scoring system be able to deal with that? No concrete evidence exists to support the idea that any facial recognition system can reliably "see" through a beard, or always accurately create a face with or without one.. If we consider that our freedom or even our very lives can be at stake, 100% reliable recognition is not a luxury but a requirement. Consider human beings who often don't recognize someone they have known all their life after they grow a heavy beard. Now, we know that humans know more about people than their Creator, right? Some companies think so. One facial recognition company uses the following grid system and claims a match as shown below
Now what about our look-alike man we talked about earlier * like the one who looks exactly like actor Kevin Spacey? (No relationship to the man in the photo above.) Facial recognition software would probably claim it found a match for him, too. We can readily see by the coarse system grid employed above that there is insufficient resolution to pick up the very small subtleties which are common between two completely different people. If the man above was dehydrated or not feeling well for example, his cheeks and neck can become somewhat depressed, distorting the imaging grid and preventing identification. Should we get our motor vehicle picture taken when we are really sick instead? We'll explore that further later.What are the other potential problems resulting from wrong identification no one talks about? Consider a security camera which captures the image of a man robbing a convenience store. He is not only stupid enough to rob the store, but instead of running off he also doesn't wear a mask. During the robbery he kills the clerk, which automatically elevates him to the FBI's "armed and dangerous" status. In society today, police won't hesitate to shoot and kill such fugitives. At this point the police computer system has the criminal's face but no name to go with it. Officers could then enter the captured image into facial scanning software, which they have been brainwashed into believing that it is reliable and trouble free. Even worse, this fugitive just happens to look EXACTLY LIKE YOU. You're at home having a peaceful evening, when suddenly your door is kicked in by the storm troopers - then you are dragged off to be charged and interrogated. How did all this happen? You can thank that mug shot on your driver's license. Whether you know it or not, that one photo of you which is now in countless government and state databases, has comprised your security for decades to come. And what about having your driver's license photo taken when you're sick and look terrible? Like handwriting, everyone's face can easily change with stress, changing of the seasons, where we move to and for other reasons like diet and illness. Even stress is a factor. You will never know until it's too late that just because you were recovering from the flu or other illness when you had your driver's license photo taken - that unfortunately you now look exactly like a wanted serial killer. And the final question is - how can you ever prove your innocent?
Reply With Quote
:
Originally Posted by Oski
Bookmarks