Multi-Modal Video Summarization

Abstract

Most previous work on summarization targeted either only on video or on its description of single video document. To efficiently manage huge video documents, this paper addresses a new prototype for summarizing video documents that automatically processes text and key-frames from the given video documents, indexes and provides mechanism for search and retrieval. The system separately summarizes the text and video related information. Both summarizations will enhance the performance as well the quality of the video document retrieval processes. The text summarizer provides a summarized view of a large video document to improve the learning speed and to give a glimpse of the document for better understanding. The video summarization consists of the key frames that effectively express the content and context of the video. This is achieved by quantifying visual attention of each frame in a video. This framework provides meaningful two-way summarization for video documents.