Understanding how visual context influences multimedia content analysis problems

Abstract

The importance of context in modern multimedia computing applications is widely acknowledged and has become a major topic of interest in multimedia content analysis systems. In this paper we focus on visual context, tackling it from the scope of its utilization within the above framework. We present a brief review of visual context modeling methods and identify and discriminate its useful types within multimedia applications, envisioning possible usage scenarios for contextual information. Finally, a representation of visual context modeling is reviewed, being suitable for aiding in the case of common multimedia analysis problems, such as object detection and scene classification. © 2012 Springer-Verlag.