Generating Football Video Summery Using News Article
Naoki Fukino, Qiang Ma, Kazutoshi Sumiya, Katsumi Tanaka · 2003
In this paper, we propose a method that summarize video. Especially, we target at football video. In this process, we don't generate summary by video picture, video audio and metadata, but we use news article which is advanced summary human generated and is easily got from website and so on. This process begin by separate play-by-play comments from announcer's comments, and associate explaining comments with play-by-play comments. Next, it estimate event section from play-by-play comments and retrieve these events using keywords in news articles. Next, it rate many video clip using keyword's similarity and time interval between event and event and so on, and decide video clip which is appropriate to each parts of news articles. Final summarized video is presented using article structure which is decided from correspondence between each parts of news article and video