Motion video annotation and analysis: an overview

Abstract

Digital storage of images and videos requires staggering amounts of storage. A mere hour long digitized video can occupy a staggering 30GB of storage. With the increasing dependence of a variety of applications on digital storage, the need for automatic and semi-automatic annotation and analysis tools for intelligent video data indexing and manipulation will be strongly felt. Manual annotation of this data will be practically infeasible and insipid. This paper presents highlights of some of the important problems to be solved for automating the process. Computer Vision techniques with a human-in-the-loop approach can provide a powerful workbench of algorithm and techniques for the task.