An Efficient Video Series Decision Algorithm Based on Actor-Face Recognition

Abstract

It is an important task to check whether or not an uploaded video on the Internet violates the copyright. One of the most common approaches for checking such videos is to compare an uploaded video with respect to the sample data provided the original owner. A disadvantage of this approach is that a large amount of sample data must be stored in the database before, and also the task becomes very challenging when an uploaded video is distorted. We propose a method of generating a signature set that characterizes a series of video sets based on face recognition data, and suggest an efficient method on how to identify which series is included in the corresponding series set. We demonstrate that the proposed method can effectively determine whether or not a video is in a series set even if the video data is relatively small as an input for the previous method.