Here’s a nifty idea: create an image search tool that lets you upload a pic and see if it’s out there on the web. I had this idea years ago, but unfortunately I’m not a programmer, a developer, an investment capitalist, or diligent. Still, here it is: TinEye. It’s technically only a beta, but you can log in and play with it.

Here’s the catch: it doesn’t work that well yet. I searched for some of my own images that I knew for a fact were out there (because I had put them on this site, Facebook, and Craigslist) as well as someone else’s images that I thought might be out there. No dice.

On the other hand, this image of the Purple Rain album cover yielded 239 results. That may not sound like much — after all, you can get 2.79 million hits on Google images just by typing in “prince purple rain.” But keep in mind that TinEye actually correctly identified the album cover in 239 different JPEGs, at different resolutions, from different scanners or digital sources, with no verbal cues whatsoever. (The file I uploaded was called “images.jpeg,” so TinEye couldn’t cheat and use the name as a clue.) The image recognition software, then, seems really solid. I’m guessing their spiders just haven’t cached enough pages yet to make it as functional as one of the big-name search engines. But keep your eye on this — if Google doesn’t co-opt it, they really aren’t evil.

This entry was posted in web 3.0. Bookmark the permalink.