CONTENT BASED VIDEO RETRIEVAL SYSTEMS

Abstract

With the advancement in technology and availability of cameras even in mobile phones, there has been a tremendous rise in the multi-media content online as well as offline. A proper database management system is need of the hour. Apart from the efficient indexing schemes there has to be a proper retrieval system for videos. YouTube is a very large repository of videos on which a user searches by text. So a video has to be tagged with an appropriate text. But there may be situations where we don’t have to search the database just by using the content of a video, where we would need to find videos similar to any given video. So ours is an attempt to develop such a model in which given a video as a query it returns as an output the category it belongs to which can further used to find similar videos. And hence reducing the time required for such a heavy search. We have basically used keyframes as our basic unit to develop a content based video retrieval system and used a supervised learning algorithm, SVM, to classify our data