Automatic Annotation of Sports Events Applying Mpeg-7 Descriptors

The amount of multimedia data available worldwide is increasing every day. There is a need to annotate multimedia data in order to allow universal content access and to provide content based search-and-retrieval. We present a tool for unsupervised annotation of sports videos, that are distributed via a video-on-demand streaming server. Our system generates a description that can be used for direct access as well as for scene based search-and-retrieval. Since supervised video annotation can be time consuming, an automatic solution is appreciated. We apply different MPEG-7 visual descriptors in order to split the video-stream into its separate scenes, where each scene is characterized by a set of visual descriptors which can be subsequently used in order to generate content summaries and to provide shot based navigation and search-and-retrieval functionalities.