SYSTEM AND METHOD FOR AUTOMATICALLY RECREATING PERSONAL MEDIA THROUGH FUSION OF MULTIMODAL FEATURES

摘要

Provided are a system and method for automatically recreating personal media through fusion of multimodal features. The system includes a multimodal fusion analyzer configured to analyze semantics of personal media having various forms based on a plurality of modalities and divide the personal media into media fragments which are the smallest units having semantics, a semantic-based intelligent retriever configured to store and retrieve the divided media fragments by considering the semantics, a personal media recommender configured to learn and analyze a profile of a user through modeling the user, and select and recommend a plurality of media fragments wanted by the user among the media fragments retrieved by the semantic-based intelligent retriever, and a personal media creator configured to create new personal media using the plurality of media fragments recommended by the personal media recommender according to a scenario input by the user.