Last Thursday at a web conference in Beijing, Google introduced the latest step in visual search technology, VisualRank.

Currently, image search delivers results based on the text associated with each picture rather than anything about the image itself. Theoretically, if somebody labeled a ham sandwich as the Taj Mahal, your image search of “Taj Mahal” would show you a sandwich. Google wants to eliminate this method by an analysis of many of the more popular images on the web.

Due to the massive amount of computer power it takes to scan and recognize a large number of images, Google and to put a good deal of resources into getting to this point. The question is, will it ultimately have any effect on the visual search sphere going forward?

The presentation was made by two Google researchers, Shumeet Baluja and Yushi Jing.

“We wanted to incorporate all of the stuff that is happening in computer vision and put it in a Web framework,” said Baluja.

The researchers concentrated on the 2000 most popular product queries on Google, items like iPods and and XBox. According to Google, the results using the new image algorithm delivered 83% more accurate results than with Google’s current text search tool.

The move towards the semantic web and and more intelligent image-based is slowly growing. While the current keyword-based search of engines like Google is not going anywhere, as the web continues to evolve and the way we find things will continue to become more exact.

Thanks to John Markoff of the New York Times for the story.

This entry was posted on Monday, April 28th, 2008 at 12:31 pmand is filed under Search Engine Marketing, Web Development. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.