Image-based search with TinEye

May 13, 2008 | 09:45

Tags: #hash #image #search #search-engine

Companies: #google

If you've ever been searching for a particular image, you'll know how frustrating the process can be. You wrack your brain for likely keywords, only to get a picture of a bunny with a pancake on its head. Wouldn't it be handy if you could just upload the closest equivalent image you have and let the power of the web do its magic thing?

The guys over at WebWare have been able to get in to the closed beta for an image-based search engine called TinEye. Built by Idée Incorporated, the engine promises an easier, more accurate way to search for images than traditional keyword-based methods.

The idea is that you upload a copy of an image you already have on your PC – perhaps it's a low-resolution version and you're after the original, or you're looking for other parts of a series of images – and the TinEye software breaks it down into a mathematical hash it can compare to images already in its database.

Idée's co-founder and CEO Leila Boujnane described the service as allowing users “to search over 487 million images in mere seconds,” and promised that the technology will “enable millions of users worldwide to search for images like never before.

Fellow co-founder and CTO Paul Bloore describes the technology behind TinEye as comprising “a number of key breakthroughs in image search that go well beyond what anyone has done before.

Despite being in closed beta, the service is already pretty polished – there's even a plugin for Firefox available which allows you to add a link to a TinEye search on the right-click menu for any image. The engine is rather limited in the sites it indexes at the moment, but as more users start using the service that can be expected to change pretty rapidly.

The technology behind the engine has already been spun off by Idée as PixID, aimed at helping content owners monitor usage of their images and track down people posting copyright material without permission.

The company behind the engine is looking toward the future, too: the next step, once the basic image search has proven a success, is to add video search which will allow for the tracking of video content even when posted in differing formats and when heavily compressed.

If you want to give the engine a trial run, you can request entry into the closed beta on the TinEye website, but be aware you may be in for a wait – all the initial places have sadly gone. No, I didn't get one – worst luck.

Whether the search engine as a tool for end-users will ever take off, or whether the technology is destined to become merely a tool for copyright enforcement remains to be seen. It's certainly an interesting shake-up in an otherwise stagnant search market, and I'm sure the competition are watching developments avidly.

Can you see yourself using an image-based search engine on a daily basis, or is it more of a curiosity item? Share your thoughts over in the forums.
Discuss this in the forums
YouTube logo
MSI MPG Velox 100R Chassis Review

October 14 2021 | 15:04