APPARATUS AND METHOD FOR GENERATING VIDEO SUMMARY THROUGH RECURSIVE GRAPH MODELING

摘要

The present invention may provide an apparatus and method for generating a video summary, in which, by considering a plurality of frames of video images as nodes of a relationship graph and inferring semantic similarity between the plurality of frames by using a graph convolution neural network having a recursive inference structure, an accurate key frame in consideration of a global long-term correlation between the plurality of frames can be extracted, thereby generating a video summary in various video platforms with a relatively small number of parameters and high accuracy.