Saturday, November 14, 2009

New Image Search: No Training Required

Engineers at the UCSC have developed a new approach to computer recognition or categorization of images or videos. They claim to have overcome a major drawback of existing methods for computer recognition of objects in images—the need for an extensive training phase using a large number of samples. With a single photograph or video clip as a template, their software supposedly sifts through thousands of images or videos and presents the ones that look most like the template. Submitted to IEEE Trans. on Pattern Analysis and Machine Intelligence.