A Video Database Management System

We develop a Web-based video database management system, named BilVideo, providing an integrated support for both spatio-temporal and semantic (keyword, event/activity and category-based) queries on video data. In this paper, we introduce the architecture of BilVideo and its SQL-like textual video query language that has the capability to handle a broad range of spatio-temporal queries, which may contain any combination of directional, topological, 3Drelation, object-appearance, trajectory-projection and similarity-based object-trajectory conditions. Furthermore, we present the Web interface of BilVideo and its tools, FactExtractor and Video-Annotator, which are used to populate the facts-base and feature database of the system to support spatio-temporal and semantic video queries, respectively. The goal of our project is to provide support for any application with spatio-temporal and semantic video query requirements; hence, BilVideo is application-independent, but yet can easily be tailored for specific needs of such applications through the definition of external predicates.