Enhancing Semantic Video Adaptation Speed through Compressed Domain Adaptation

The delivery of videos to the increasing number of heterogeneous mobile devices over different types of networks necessitates video personalization. This task should be based on device and network capabilities as well as semantics of the video as perceived by the user. Personalization of videos based on semantic user preferences can be computationally expensive and affect the speed of video delivery. In this position paper, we present compressed domain semantic video adaptation as a means to reduce the computational cost of video personalization.