DAVOS, SWITZERLAND: Swedish start-up Polar Rose plans to make its face recognition service -which it claims can sort and index photographs according to faces – publicly available on the Web in the second quarter of 2008, as it tries to become a must-have tool for sorting visual content. The firm says its technology can pick out faces in the swelling crowd of images on the Internet, thereby making digital photos indexable – just like text documents. Polar Rose’s free software for making photos searchable is available as a browser plug-in, and soon will also be embedded on partner Web sites. It aims to integrate the system on the first partner sites next month. Existing search functions generally find images by scanning text attached to pictures. That system, however, does not work if the text ‘tagging’ is wrong or absent. Polar Rose’s technology scans the image itself, and converts the data from two-dimensional (2D) images into 3D models. These skeletal models can be rendered into so-called ‘faceprints’, which are then stored and indexed. The technology should allow users of services like online photo-sharing site Flickr to sort and group personal photos face-by-face. More broadly, it will let people find similar-looking photos across the Web. With the number of images on the Web doubling every seven to eight months, the company is confident its know-how will win backing from advertisers. The system has been in beta testing since last July. However, the small outfit is not alone, with Google having bought rival photo-recognition firm Neven Vision 18 months ago. But Polar Rose’s Chief Executive Nikolaj Nyholm thinks his company has an edge. "They (Google) have some raw power we can’t really match. But from a core technology standpoint, we feel we can still do better matching than (their) technology," he said. REUTERS
NO USER COMMENTED IN THIS POST