Today, Idee is taking the image-recognition chops it has built up through corporate image searches and applying them to consumer-level searchs through a beta called TinEye.com. Using either an image from your hard drive or a link to one on the Web (the service also has a Firefox plugin that adds TinEye to the right-click menu), the service can almost instantly produce a list of similar images — even when the image in question has been stretched, shrunk, cropped, flipped, had the colour profile changed, or been otherwise modified.
I saw a demo of the corporate version of this technology a while back and was blown away, and now it is being made available to anyone. Unlike most image-search services, which use the text and keywords associated with a photo or image, Idee uses the digital “fingerprint” of the actual pixels in the image and compares that with others until it finds a match. Other companies have claimed to be able to do this in the past, (including Riya, which then became Like.com, a shopping search engine) but none have impressed me as much as TinEye.
There’s more info on TinEye at the Idee site (including the fact that it is currently crawling almost half a billion images), and there’s also a very helpful video explaining the service that features another friend — Amber MacArthur, video-blogger extraordinaire and host of CommandN. I’ve got a limited number of invitations for the service available: drop your email into a comment or use my contact form (link in the upper right-hand corner of this page) and I’ll hand them out on a first-come, first-serve basis. Congrats to Leila and the rest of Idee.