Elliott C. Back: Internet & Technology

Google Image Labeler

Posted in Google,Interface,Web 2.0 by Elliott Back on September 1st, 2006.

Google’s new Image Labeler product tries to turn around the way it classifies images by letting pairs of people team up to tag them. As you enter tags for an image, Google compares your tags with your anonymous partner’s tags and selects the first match. The interface works well enough, but is a bit too simple for my taste:


This new product hasn’t affected the stock price of GOOG:

When you finish categorizing 5 images, for which you receive 100 points each, Google Image Labeler lets you know, “Thanks for your contribution. It will help us improve the relevance of image search results so that you and other Google users can quickly and easily find the results you’re looking for.” Every 90 seconds you can tag an image, so if 1000 people are tagging images at the same time, say with an average of 60s per image, you can tag 5000 images in no time at all. When algorithms and computers can’t do the job, use people.

My final thoughts are that Google is desperate in the field of image search and feels like it has exhausted all its possibilities. Because it can’t make computational headway in image search, it’s turned as a last resort to using humans to process its images. Compare this to startup Riya, which is probably the only image search innovator I can think of right now…

This entry was posted on Friday, September 1st, 2006 at 6:34 pm and is tagged with . You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback.

3 Responses to “Google Image Labeler”

  1. Will says:

    That might mean you don’t have as much time to kill /: ) I’ll be the first to admit that I consider it a good use of my time.

  2. Yeah, they did aquire Neven Vision, has anyone else used this Image Labeler? I did and I wasnt really impressed, although it seeems like a great idea, it still pretty bland.

  3. Will says:

    Although it seems simple, there seems to be some academic underpinnings. The help page says the app was developed with technology licensed from CMU. Although they don’t have Riya, Google aquired another company that has very similar technologies even broader in scope than facial recognition. Do a search for neven vision… I’m not risking any links.

Leave a Reply