| 摘要 |
A system for recommending a section of interest from a video according to an embodiment of the present invention comprises: a user information processing unit which, when a video is played, recognizes at least one among an expression and biological information of a user to determine whether the user is interested in scenes in the video, and extracts playback time values for the scenes of interest, which attract the user′s interest, from the video; a log data processing unit which extracts log data including information on an object from each of the scenes of interest; and a learning processing unit which extracts a characteristic vector for the object from the log data and generates learning data for the scene of interest based on the extracted characteristic vector and the playback time values. According to an embodiment of the present invention, it is possible to learn a model using a deep learning algorithm by utilizing big data which is accumulated as the user consumes video content. |
