faces
Each time I re-start my blog or re-design my site, I usually look over my old content and decide to throw it out. Despite my best intentions at the time, I always discover that my past writing is laden with ignorance and sullen tripe about Cindy Lou Girlfriend and the highs and lows she had inflicted upon me. No one wants to read that.
Since graduating from college, I've tried to keep my writing more interesting from the perspective of Future Dan, and I think I've done a decent job (not great, but decent). My last blog was mostly free of the aforementioned tragedies, but I didn't include it in my latest re-design because I was too lazy to refactor those posts whose content did not fit well with the new layout. It was just easier to start over. Fortunately, I saved a few of my favorite posts from that blog, and that brings us to today's post.
I don't want to waste a lot of posts (or even multiple posts) on old material, but I was reading an article the other day that related to an interesting topic that I had previously covered in great detail. The article was about a Carnegie Mellon research project that demonstrated success in patching incomplete or damaged images using a catalog of disparate image fragments found online; this news represents another step in the long journey towards accurate and accessible image search that does not rely on pre-existing metadata or titles[1]. Very cool stuff. It also has potential with regards to today's topic, which is the patterns of the human face and the way the brain processes those patterns.
Here is what I wrote a few years ago on the topic of facial patterns:
I think it would be really interesting if someone created an ontology of the human faces for each generation. It is my belief that there is a finite number of possible faces, and that, unlike snowflakes, we have a lot more look-a-likes than people care to admit. A comprehensive face catalog would tell us exactly how many unique faces are out there. I think that we would all be astonished at how small the number is.Note the last paragraph: the real challenge would be providing a mechanism for efficient lookups. Well, science marches on. Projects like the one at Carnegie Mellon continue to chip away at the seemingly impossible task of finding images based on natural language descriptions or imperfect renderings (such as sketches, or photos of things that are only conceptually similar to what you are looking for). Within a few decades, we may be able to study the American facial spectrum without the manual and tedious processes I imagined when I wrote my original post. My only hope is that it will be used for good[2].
Here's an experiment for you to try: gather a high school yearbook from a friend or family member that you did not go to school with, and look through the senior photos (which are always larger and more detailed). I guarantee that you will recognize numerous people despite the fact that you don't know them. You will find photos of people that look exactly like people you went to school with, met at a bar, etc; furthermore, you will find faces that you've seen on multiple people, in their exact form.
Straight-haired brunette, thin face, pale complexion, wears a little too much makeup, squints too much when she smiles, mouth is thin as a piece of paper. Square-faced, black hair with too much gel, clear blue eyes, chubby nose, has a grin that makes him seem uneasy. I'm telling you, the similarities are there. There are very few people in this country that I have not "met".
This does not mean that people who have popular faces are mundane or unattractive; if you do the yearbook experiment, you will find a lot of gorgeous people that happen to look just like some other gorgeous people you know. My hypotheses are simply that a) a significant percentage of the population has more twins than they could ever imagine, and b) this is the exact opposite of what most people believe.
And that brings me to my social circle. I don't think that I look like a significant number of other people, but then again, there is nothing very remarkable about my appearance. Mike, who agreed with my theory when I related it to him this afternoon, is also an uncommon face. Or not. It's not very likely that my small circle of friends has somehow escaped this phenomenon. This leads me to my final theory: c) even people with a twin in every city don't realize that they have a very common face.
The only time I see exceptions to my theories is when someone happens to look very similar to a current celebrity. I am guessing this is due to the fact that our brains are able to put a name to the famous face and more easily remember it during day-to-day life.
Now, what's really interesting is that the types of faces seem to change from generation to generation. My analysis might be skewed by different fashions, trends, etc., but it seems like facial patterns are limited to two decades or so. What I would like to see is an American face ontology for my generation; the real challenge would be providing a mechanism for efficient lookups. If that were possible, you could categorize any of your friends as, say, a #52, which is both amusing and scary at the same time.
[1] For an example of some really innovative research in this area, check out Retrievr, which allows you to search images on Flickr by drawing sketches or uploading photos with similar content. Make sure you have JavaScript enabled when visiting the site.
[2] A more likely scenario: two days after such a project is announced, there will be a Firefox plugin that lets you stalk people you knew in high school and find out if they're single and miserable.
Labels: culture, narcissism
0 Comments:
Post a Comment
<< Home