This paper studies usage of both keywords and captions in one scene for video content. Captions show the spoken content and are renewed in a sentence unit. A method is proposed to extract keywords automatically from transcribed texts. The method estimates topic boundary, extracts keywords by Latent Dirichlet Allocation (LDA) and presents them in speech balloon captioning system. The proposed method is evaluated by experiments from the viewpoint of easy to view and helpfulness to understand the video content. Adding keywords and captions obtained favorable scores by subjective assessments.
|出版ステータス||Published - 2010|
|イベント||2nd Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2010 - Biopolis, Singapore|
継続期間: 2010 12月 14 → 2010 12月 17
|Conference||2nd Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2010|
|Period||10/12/14 → 10/12/17|
ASJC Scopus subject areas
- コンピュータ ネットワークおよび通信