I figured I would post something about this because I haven’t seen it discussed on any forum I am familiar with on the Internet.
Facebook reveals a little information about the information it collects from the pictures you post in the form of ALT tags. ALT tags are included in the raw HTML code of nearly every website you visit. They are primarily used to briefly describe a particular picture, image, or graphic for users that can’t view images, either because they are visually impaired, or their Web browser doesn’t support it.
By inspecting the HTML code of anyone’s Facebook profile, you can see some of what Facebook can recognize in your photos. Take for instance these examples on my own profile. In the first example on the thumbnail with my girlfriend and I in the car, Facebook believes it sees 2 people, smiling, sitting, outdoors, with sunglasses and a beard. This is amazingly accurate (it HAD been a few days since I shaved!) In the second example on the thumbnail of me jumping my four-wheeler at Jordan Supercross, Facebook believes it sees one or more people on a motorcycle outdoors. Again, not too shabby on Facebook’s part!
If you care to experiment a little, what other criteria can you find that Facebook’s AI can successfully identify? Share here!
