VIRON: an annotation-based video information retrieval system

To provide efficient search and retrieval of video data from large archives, we need to model video data appropriately. We propose a video data model and describe the design and implementation of the annotation-based video retrieval system VIRON (Video Information Retrieval On Notation) based on the proposed model. This model provides the mechanism of sharing and reusing annotations among users by introducing descriptor schema. In order to process a query efficiently, an annotated video unit is mapped into a unified video annotation stream. VIRON is composed of three tools: CVU manager, which is used to manage and visualize conceptual video units; annotator to annotate video data by offering an interactive video player; and a video query tool to pose and process video queries.