映像・音声認識,自然言語処理の適用によるメタデータ生成の作業コスト削減効果に関する考察

概要

論文の詳細を見る
We propose a task model that semi-automatically generates scene-based metadata based on media analysis technology such as audio/visual indexing and natural-language processing to reduce the costs of generating metadata. Our task model can shorten the task time by reusing both the results of media analysis and existing text information such as program scripts. Scene Cabinet, a metadata generation and editing system, can automatically extract scene-based metadata from videos. The system extracts meaningful video slices and textual information such as scene titles, synopses, and keywords using natural-language processing based on the results of speech recognition and video OCR. Moreover, the system can import program scripts and use them to automatically extract keywords. Scene Cabinet provides an intuitive user operation interface including a video browser with key images that are automatically detected based on scene changes, on-screen text, camerawork, speech, and music. Experiments showed that Scene Cabinet could significantly reduce metadata generation costs.
社団法人映像情報メディア学会の論文
2007-06-01