1. Immersive video: The proliferation of head-mounted display devices capable of rendering in real-time virtual 3D scenes may pave the path for the delivery of actual immersive content. The research challenges here lie in the acquisition, representation, transmission, and reconstruction of such content. For example, the advent of image capturing technologies means that the dimension of media data at a server (e.g., light field, hyperspectral imaging) can far exceed the dimension of the output image display (e.g., one 2D image at a time on a conventional monitor) for human observation. Clearly transmitting the entire volume of high-dimensional data to the client when only a small data subset is visualized is inefficient. The research challenge is thus how to structure the coding, extraction and streaming of high-dimensional media data in such a way that the client can seamlessly and interactively navigate the media space, requesting one media data subset at a time for correct decoding and rendering.
2. Real-time in-network big media analytics: the ubiquity of surveillance cameras and popularity of on-demand videos (e.g., Netflix, Hulu) have made video the “biggest big data” in today’s packet networks. To maintain system scalability despite the video data explosion, processing media in-network in real-time is essential so that only important data are extracted and summarized while extraneous data are filtered. The research challenge is thus to enable application-specific in-network media filtering to maintain system scalability, while preserving data security, personal privacy and deployability.