An indexing and browsing system for home video

In this paper, we present a content-based video indexing and browsing system for home video. In addition to traditional video parsing to extract temporary structure and abstraction of a video, we also utilize face detection, tracking and recognition techniques to create an organization based on the existence of people in a video. The entire parsing process is designed in the way that it takes full advantage of the information in MPEG compressed data so that it can be performed very efficiently. The system also provides semi-automatic tools for semantic annotation and fully automatic content-based retrieval.