Agile Views for Video Browsing: Advanced Surrogates, Control Mechanisms, and Usability

University Of North Carolina At Chapel Hill, Chapel Hill NC

Investigators

Abstract

This is the first year funding of a three year continuing award. This project will afford people agile views - effortless control over different representations - for digital video objects, by providing a multiplicity of indexes that are attuned to people's experiences, leverage the features of the content, and are easily and rapidly used and changed. The PI will specify such an interface, create prototype instantiations, and develop and apply procedures for testing the usability of the prototypes and specifications. Most research on digital video systems focuses on the retrieval of specific objects and on one or a few indexing attributes. This project is unique in three ways. First, it aims to address tasks at the collection level as well as at the item level; thus, in addition to the retrieval task, it aims to help people understand a video collection's structure, what is and is not available, and what attributes might be useful for retrieval purposes. Second, it aims to provide people with a range of surrogates and to integrate these into an effective and efficient interface; it aims to create an environment that provides multiple surrogates at the collection level as well as at the individual item level, and to provide novel control mechanisms to manage these alternatives. Third, it aims to assess user performance on these different tasks using these surrogates; evaluating browsing behavior is a challenge in any medium, and this work will build upon tasks and metrics developed in previous studies to go beyond traditional metrics to address time-benefit tradeoff measures. For this project, the video objects will be drawn from the Open Video Repository. Surrogates such as key-frames (in slide shows, storyboards, and skims), audio extracts, and keywords will be used as the basic representa-tions for specific video segments, and will be user-manipulable through a variety of interaction mechanisms. Additional surrogates (such as two-dimensional layouts of metadata and coordinated metadata lists) for collections of videos will also be developed, and additional views, i.e., histories (reviews), peripheral views, and shared views, will also be investigated within the agile views environment. The usability of the agile views environment (its individual components and the completely-integrated interface) will be evaluated iteratively, concluding with assessments of user performance (object and action recognition, and video comprehension), tradeoffs associated with viewing compaction rates, and user satisfaction with the interface. These evaluation techniques will themselves contribute to research and development by providing metrics and techniques for assessing interactive browsing. The ultimate goal of this work is to provide an information-rich and interactive environment that enables people to go beyond their innate visual and audio abilities to browse video content and process large volumes of video information.