Automatic Movie Summarization Based on the Visual-Audio Features

Automatic movie summarization based on the visual-audio features is introduced. The importance evaluation model for video clips chooses face detection and atwitter shot detection as the basis. Against the difficult of movie's speech endpoint detection, speech endpoint detection based on subtitles of movie is introduced, which can help detect the captions' start or end time exactly. Compared with other summarization algorithm, the introduced algorithm is suitable for movie summarization.