The science of recognizing a face and the intelligence to recognize the same one twice. Or a million times. That’s one of 2009’s crowning technological achievements. Oh, sure, there have been facial recognition algorithms around for years, but up until this year, we haven’t had that feature on a mainstream consumer level. Within the past year, we’ve seen point-and-shoot digital cameras receive the ability to focus on individual faces in a frame and some of those cameras can even detect whether or not you’re smiling (and summarily refuse to snap the photo!). Standalone desktop applications like Apple iPhoto and Google Picasa will now scan your entire photo library for faces and once you’ve given the software some training, it’ll auto-detect and tag the mugs of family, friends, and maybe even your dog. In fact, Google’s online photo catalog software, Picasa Web Albums, has been doing the same thing for awhile as well. It’s a feature that makes cataloging photos much, much easier. No longer must you rely on finding an exact folder or using a non-smart desktop search tool to find the person you’re looking for. As long as you can remember their name, you can find their face.

For as long as most of us can remember, Facebook has allowed photo tagging. Upload your photos, click a cross-hair across a person’s face, and type in their name. If they’re using Facebook, it’ll link directly to their profile page and notify them that their face is now floating around the ‘net for any of their friends to see. A cool feature, very informative, and a stalker’s dream. But one thing that’s been bugging me for awhile is: why stop there?

Facebook, no doubt, has tens of petabytes worth of data stored across its servers and much of that is photo related. People upload terabytes worth of photos every day. Why isn’t Facebook using all that meta information to make your life easier? Imagine uploading 75 pictures of you and your friends from last week’s party. Facebook would then analyze each photo, tag everyone it recognizes from your friends list, and give you a confirmation page to adjust any of the tags. My guess is that the system wouldn’t even require training from the user as it could pull aggregate data from profile pictures and other tagged photos in order to make the whole process incredibly quick and easy. The question isn’t if they could do this, it’s why have they not?

Two reasons come immediately to mind, the first being that of privacy concerns. Suddenly you’ve turned Facebook into a huge searchable photo database that could make any law enforcement officer’s day by providing thousands of images and demographics of people not available in government or criminal databases. You’d also be presenting yourself as a target for some dangerous information leakage if Facebook were ever breached. This concern, I believe, is mostly without a logical base. Users continue to manually tag friends anyway and unless you disable the tagging system altogether, privacy will always be a concern.

The second potential issue is one of raw processing power. For instance, it took my beefy ThinkPad about 4 hours to sort through my collection of approximately 25,000 photos. That’s one dual core CPU’s near best effort since the average processor utilization was somewhere around 70%. I’m guessing Picasa was leaving some of the CPU free so that the machine didn’t slow to a complete crawl. Imagine then the billions of photos stored across Facebook servers. The company would need to create a system to process every single photo containing a face, match already associated tags, and create a massive database containing the results. This process would require not only a lot of time, but much additional storage space as well. Fortunately, storage is cheap these days–unfortunately the required number of extra CPUs and RAM modules are not.

As a side note, there was a Facebook Application released in the not-too-distant past allowing users to scan their photos for faces, but my guess is that without widespread use (and some major venture capital), the corresponding company and software didn’t get too far.

Ultimately, I think facial recognition is something that Facebook should and will eventually add to it’s service. My guess is that this feature is already in the works; however, Facebook usually holds close its cards until they’re ready for some sort of official release plan. Keep watching Facebook for any new information, and of course, you’ll find the details here as well when the announcement does arrive. My prediction? We’ll see something emerge by the end of 2010.

Cross your fingers…