Video Summarization using Neural Networks

Abstract

With the rapid increase in video data, the need for video summarization has increased. it allows us to search and retrieve video data easily. In this work, we have tried to solve the problem of video summary using unsupervised learning. Our method is based on the extraction of key frames. So we rst segmented the shots and extracted key frames from each shot. And these key frames are combined to generate a storyboard. We have experimented with di erent clustering techniques such as k-mean, k-medoid and self-organizing maps. We evaluated our results using a human-generated summary.