Abstract: Automatic visual information retrieval methods are typically easy to use but have well-known issues, such as the semantic gap or the usability gap. Pure navigation-based methods on the other hand are usually too tedious to use, especially for large multimedia data archives. In this talk I will discuss visual content browsing, a content-based exploratory search approach that integrates methods of automatic retrieval and interactive search. I will explain how visual content browsing can help us to overcome the shortcomings of automatic image and video retrieval. Special emphasis will be put on the interactive part, although content analysis methods will be introduced as well. Furthermore, I will talk about the evaluation of visual content browsing tools, mention related issues and introduce the Video Browser Showdown, which is an annual competition to evaluate such tools. The talk will be concluded with an outlook of future opportunities and challenges of visual content browsing.
Short bio: Klaus Schöffmann is assistant professor at the Institute of Information Technology at Alpen-Adria-Universität Klagenfurt, where he works in the Distributed Multimedia Systems group (Prof. Laszlo Böszörmenyi). He holds a PhD in computer science, for his work on Immediate Video Exploration. His current research focuses on visual content analysis, multimedia information retrieval, multimedia interaction and content visualization.