During the International World Wide Web Conference in Beijing, NYT  reported that Google has now begun testing a new algorithm that would perform exactly the same functions for images as the PageRank does for the text of a webpage. That means, it will be scanning images on the web for Image Ranking. This technology is known as ‘VisualRank’, which is a blend of image recognition softwares combined with other techniques to rank images.

Inspite of the fact that the images are richer, vibrant and popular, still all major search engines rank webpages on the basis of the text rather than the images that are present on webpage and that make the page more presentable. According to Shumeet Baluja (Senior Staff Researcher, Google) said, “We wanted to incorporate all of the stuff that is happening in computer vision and put it in a Web framework.”

During the testing period, Google has tested 2000 product queries on Google’s product search. Keywords and names like iPod, Xbox and Zune. After the test, top 10 ten images from the image search as well as from Google search were sorted out. With a team of over a 100 employees, a scoring system was conceived for ‘relevant’ images. The overall return was a spectacular 83% less irrelevant images.

Having said that, Google isn’t the only company that have tried their hands on visual product search category, image ranking. In 2006, a Silicon Valley based IT company known as Riya introduced Like.com. The service was that of referral category, where users were referred to shopping websites, thereby enabling users to select a particular visual attribute, on the basis of which the user is then presented with similar products from competing online companies.

This entry was posted on Thursday, May 1st, 2008 at 5:11 am and is filed under Search Optimization. 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.