MACHINE LEARNING ALGORITHMS, DESIGN & IMPLEMENTATION DETAILS IN SMART TELEVISION BY IMAGE AND MULTIMEDIA PROCESSING

Abstract

A solution for better view ability and audibility features for Smart Television has been proposed in this thesis which will provide better and enhanced technological benefits to end user (Television Consumer); moreover, it will provide an opportunity to Smart Television manufacturers to implement machine learning algorithms, which is not possible in current situation because Multimedia stream specific decoding is done at Set-Top box. Since in this setup television act as a dumb rendering device; we have developed a novel approach by which television can extract intelligent information from data stream which was intended for rendering of multimedia frames pixel by pixel. Smart Television Industry will also get an opportunity to revive the consumerism which is now slowing down in industry because many of the technologies described in this thesis would not be possible only merely by software upgrade; since it requires new hardware changes which means that television manufactures have an edge to sell brand new hardware (Television) to customers which simply means more sales and can revive the slowdown that television industry is experiencing in emerging markets.