A Hierarchical Video Annotation System

Semantic search service is generally considered to include text search, audio search, and video search. The annotation of a video is crucial to the success of video search and retrieval. With the precise annotation of video files and their semantic reasoning, more accurate search results according to the users' query intention can be obtained. In this paper, a prototype of such a video annotation system, called Semantic Video Annotation System (SVAS), is presented. SVAS uses a three-level annotation architecture and a semantic video search language called Semantic Query Description Language for Video (SQDL-V). SQDL-V engine based on SVAS is able to return more accurate search results in comparison to the formal video search method.