New image recognition software methods to facilitate online image searching
Google has just revealed details of its new image recognition software methods that will be utilised for facilitating online image searching. Google scientists and researchers, speaking at the International World Wide Web Conference in China, described VisualRank - an algorithm for combining methods for ranking images which look similar, with image recognition techniques.
According to the Google experts, image analysis still remains a largely unsolved area in the domain of computer science. As looking at all indexed images would be rather impractical, they have reportedly focused more on a subset of the images catalogued in order to analyse and compare the digital images. However, the Google team members claimed their image Search to be the ‘most comprehensive image search’ available on the web.
A team of around 150 employees has been working to create a scoring system for the relevance of images that are related to these products. Google has opted to concentrate on the most popular and prolific product queries on its product search, comprising terms, such as Zune, iPod and Xbox.
Commenting on the exercise, a senior staff researcher at the search engine giant, mentioned.
We wished to incorporate all of the stuff, which is happening in computer vision and place it in a web framework.
No Comments
No comments yet.