Once you're on the page, simply upload a photo (or enter an image's URL into the appropriate box) and click the "Call Method" button. A few moments later, you'll see your image overlaid with some strange markings. Hover your mouse cursor over any of the those dots and boxes to see what Face.com's API calculated. There'll be details such as the estimated age, minimum age, maximum age, gender and mood of any individuals in the image. You'll also see whether or not the software believes that a person is wearing glasses or smiling.
The image analysis will also offer percentages after each detail. Those indicate how sure the software is about any particular number or characteristic. Many factors, such as as image quality, wrinkles, facial smoothness, poses and so on will affect both that percentage as well as the final result, according to Venture Beat's Sarah Mitroff.
I tested Face.com's new API with several images of individuals whose ages I know and found that it's fairly accurate, within a three year range, in most cases — but that it can sometimes make dramatically wrong assumptions about photos of people who are making silly faces or posing in dramatic light.
Face.com chief executive Gil Hirsch probably wouldn't be surprised about these results as he did mention to Mitroff that "[o]n average, humans are much better at detecting ages than machines."
But absolutely accurate or not, Face.com's new API is certainly a fun way to waste a couple of minutes. And who knows, maybe it'll help you finally narrow down your co-worker's age range.
Tee hee... Get yr fucking camera outta my face!