has built a search engine that finds images similar to those you upload or draw. Now the company is spinning out of Hitachi, moving its headquarters to Silicon Valley and when summer comes going into beta stage with its search engine.
The technology is based on putting different colors and shapes into groups, or clusters. Because images contain much more information than text, searching through indexed images takes more time than ordinary text search. Searching through the colors and shapes of all images on the Internet takes too long. The Gazopa technology extracts image features and clusters them according to their similarities in color and shape — red images in one group, round images in another, etc. Searches are performed only in the clusters that contain the specific color and shape of the uploaded image.
Gazopa has indexed over 50 million images and is able to deliver up search results within one second (not including upload time).
Can this search engine generate income? Well, it’s no secret that the number of images on the Internet is exploding. Image warehouse Imageshack have 20 billion images stored, Facebook has 15 billion, and Flickr has 3.4 billion, all growing fast. Facebook users are adding photos at a rate of 850 million a month. Many stored pictures are not indexed, nor copies of what is on your hard drive or Flickr account so that you can do a search for the exact copy of your image, like in the case of GazoPa competitor TinEye. Gazopa’s searches for similar images might just have found a gap in the search business.
GazoPa Project leader Hideki Kobayashi outlines two target groups for GazoPa’s search engine: stock-photo agencies and e-commerce sites. Stock-photo agencies can search efficiently without the time-consuming effort of tagging images. E-commerce customers can let customers search for products they already have an image of. GazoPa is working on three different revenue models: providing search-related ads, taking a commission when users purchase at e-commerce sites, and licensing the search engine to e-commerce sites.
The GazoPa search technology has been a 10-year project within Hitachi-Japan. The company will spin out of Hitachi in during 2009 and some of the team members are moving to Silicon Valley next month which will be the beginning of moving it’s headquarter’s to Silicon Valley and start looking for external funding.
By the way, I tried service out to see if there are any photos out there that look like me. The result GazoPa came up with? Lenin. As a statue, lying in state. I’m humbled. Apparently, I neglected to choose “face” search mode.