'What is in that video anyway?': In search of better browsing

Abstract

Effective use of digital video can be greatly improved by a combination of two techologies: computer vision for automated video analysis and information visualization for data visualization. The unstructured, spatio-temporal nature of video poses tough challenges in the extraction of semantics using fully automated techniques. In the CueVideo project, we combine these automated technologies together with a user interface designed for rapid filtering and comprehension of video content. Our interface introduces two new techniques for viewing video and builds upon existing techniques to provide synergistic views of the video content. We also report on a preliminary user study that compares the efficacy of these views in providing comprehension of video content.