Project description
Digital multimedia collections have become ubiquitous and are growing rapidly, due to the widespread use of digital cameras, mobile phones, email, and the web. They contain a wealth of information, but if and only if properly categorized. To reach effective categorization of images, we propose a visual analytics approach, tightly integrating multimedia analysis and information visualization. We base the work on various established metaphors and will extend them to multimedia categorization. Challenges are to develop analysis and visualization tools that use all data, including given metadata and automatically derived content-based metadata, in a synergetic manner; to deal with the large scale of the collection; and to bring the interacting user and the automatic processing together such that both system and user are as effective as possible. The research is unique in bringing together multimedia analysis and information visualization, the two complementary disciplines required to reach the goal; its focus on mixed media collections instead of either the metadata or the visual data; and its evaluation in practice on real use cases with expert users. We expect effective and innovative solutions for the tasks as results, as well as generic insights in visual analytics approaches to multimedia categorization.Researchers
- Marcel Worring (PI)
- Jack van Wijk (Co-PI)
- Arnold Smeulders (Scientific Advisor)
- Jan Zahalka : Large Scale Interactive Multimedia Categorization (Phd student)
- Paul van der Corput : Information Visualization for Categorization (Phd student)
