Representing Video for the 50,000,000 Channel Future|
Interval Research Corporation
It is hard to imagine our own civilization before the advent of widespread literacy in the 17th and 18th centuries. In the next century our descendants might find it hard to understand that while everyone watched movies, videos, and TV, so few made them. Technologies for representing, retrieving, and repurposing video content, like Media Streams, will enable radical changes in video production, distribution, and reuse. It may be hard to conceptualize a world in which you engage in a daily practice of making movies from parts of existing ones to communicate and play with. Media Streams is a system for representing video content that enables humans and machines to work together to annotate, browse, retrieve, and repurpose digital video. It uses a stream-based iconic visual language that represents semantic and syntactic properties of video content in order to enable the retrieval/composition of video sequences from an archive of annotated footage. Media Streams' representation and retrieval technologies employ cinematic concepts of space, time, character, and action. Its representational framework uses a hierarchially structured semantic vocabulary of composable primitives which overcome the limitations of keyword-based systems. Media Streams' iconic visual language addresses the limitations of text based technologies for reading and writing representations of video content for a global media archive. Media Streams' representational structures and retrieval algorithms enable user queries to compose new, rather than just find existing, sequences. In the near future, content representation technologies will enable video to finally become a computational medium such that every computer can be a TV station.
[ IBM home page | Almaden Research Center | IBM Research | (C) | (TM) ]