TinEye: Search the Web with images instead of words

Traditionally, Web search has relied on words or queries to scan massive indexes of pages for results. Searching for images can be a little trickier though. You're often relying on the competence of whoever uploaded the shot to provide the proper file name--and in a very small percentage of cases extra hints in the metadata to help the search tool get its hooks in.

In the real world, this isn't always the case, which is why the creators of a new image search tool called TinEye have approached image search the other way around--letting you search for sites using image files you've stored locally.

Sure this seems a little backward, but the idea is to find content related to whatever pictures you've got stored on your computer, or simply discover variations of that same shot around the Web.

The tool works best with popular or otherwise well-known images. Nearly everything else I tried didn't produce much. Some of the searches with the most results have been compiled in a "cool searches" section, but typically feature well-known art or photography. The company expects to have a better set of results as its index increases in size.

One of the more interesting uses for this technology is tracking down stolen intellectual property. We often find our posts put up on small blogs that cut out the bylines and take credit. If the blogs reused whatever screenshots we've added, a tool like TinEye would track them down even if they're using slightly altered text.

While we probably wouldn't go to such lengths to hound down a screenshot, there are photographers and other content creators who would. The company has already spun its technology off as a product called PixID, but it's geared more toward larger content rights holders than small-scale bloggers.

For users who want to spend a little less time (and bandwidth) using TinEye, there's a Firefox plug-in that will add a "search for image on TinEye" option when you right-click on any picture you come across on the Web. Users can also just grab the URL and enter it, which can be helpful if you don't feel like ferrying image files back and forth.

Several sites are already using the technology behind TinEye. Digg uses it to help search for duplicate story submissions on any post that contains an image. Adobe also uses it in PhotoShop Elements to look up related images that match the colors found in your photos.

Moving forward, the company intends to add video to its repertoire, letting you see where a video has been posted regardless of what service it's hosted on. This is something I'm far more interested in seeing than photos, as the viral spread of video clips across blogs and other pages is fairly rampant. When mixed with some sort of timeline, this service could yield some great metrics for video creators and a tracking system to follow when videos have been remixed and re-edited.

The service is currently in private beta, but we've got 100 invites for Webware readers who want to give it a spin. To get yours, fill out the Wufoo form after the break.