The Facebook photos application has been by far Facebook’s biggest success, the easy to use interface for uploading pics from your computer and then brilliant point-and-click tagging of friends has proved intensely viral; there are now more than fifteen billion photos on Facebook, making it the largest collection of pictures in human history. More than 850 million new photos are uploaded a month with no signs of slowing down, and Facebook’s servers display twenty billion photos every month.
F**k it, we'll do it live!
Our biggest ever edition of TNW Conference is fast approaching! Join 10,000 tech leaders this May in Amsterdam.
The question is; how many of those pictures are of you? Well you know how many of them you’ve been tagged in but how many photos are out there that you have no idea about? A new application launched today aims to answer that question.
Photo-Finder will go through all the albums in Facebook that your account has access to, and using their fast, powerful and accurate facial recognition technology, scan Facebook to find untagged photos and let you browse through the results. There’s no need to train Photo Finder, it does that for you. Photo Finder shows you all of the results sorted by accuracy or date, letting you review its findings so that future searches become more accurate. With Photo Finder you also get notified whenever a photo of you gets posted, even if no one tagged it. Photo-finder lets you know first and gives you the chance to hide potentially embarrassing photos from other Photo Finder users.
Photo-Finder also has the ability to push information on newly discovered photos to your FB news feed, giving this app a real chance to go viral as more and more people discover pics they never knew about.
The Technology behind this app was developed by the app’s creatorsFace.com. This patent-pending technology is based on Face.com ’s own proprietary algorithms that allow them to achieve high accuracy for recognizing faces in the real-life photos of the web on a massive scale. In an experimental real-life faces dataset (“Labeled Faces in the Wild”) organized by the University of Massachusetts, an Hybrid descriptor-based algorithm published together with the face.com team outperformed significantly best-known algorithms to date; comparison at http://vis-www.cs.umass.edu/lfw/results.html .
It seems that Photo-Finder has a very real chance of changing the way we interact with the Facebook photos platform and perhaps give us a real taste of a future where computers can actually recognize very complex visual data like faces and correctly tag them. The founders behind Face are close friends of mine and knowing their combined set of skills and technological abilities, I am certain that this application will be more technologically advanced than any other we’ve seen so far.