Regular readers of this blog know that I’m a big proponent of geotagging. As I’ve gotten deeper into Facebook, I’ve also become a big fan of people tagging.
People tagging allows you to uniquely identify people in pictures. Tagging can also be a collaborative effort. When I’ve uploaded pictures with people I don’t know, my friends have filled in the gaps.
Viewers can rollover the picture and the names are displayed.
It’s surprisingly addictive and it forms the basis of the real power of Facebook photos. More on that later.
It also raises privacy issues, beyond those that I discussed yesterday. On flickr, I deliberately don’t include last names because I don’t want the pictures to be searchable in general purpose search engine. I’ve been more comfortable uniquely identifying people in Facebook’s more closed environment.
Still, I wonder about the database that I’m helping to create. Given the enormous popularity of Facebook photos — it’s the number one photo sharing site — it’s likely that Facebook has the largest privately held database of individually identifiable pictures in the country. It would make a great training set for image recognition software. You’ve got uniquely identifiable people in a variety of situations and camera angles.
So far, most public efforts at image recognition haven’t been very successful. Riya, which started out as a visual people search tool, used tags and other metadata to help improve the results. Even that wasn’t good enough. Riya has largely refocused on identifying merchandise.