VANAHEIM is a real time application for analyzing behavior streams from surveillance cameras and suggest interesting, unusual scenes to an operator. In order to accomplish this goal we tried to simulate human attention processess and their relation to computer vision algorithms.

Attention stands as a fundamental capacity of our cognitive systems, for it enables us to effectively use
its limited resources in a flexible and dynamical way. Therefore, it is not surprising that a considerable amount of research has focused on that topic. However, most of this research has been mainly carried out on visual attention by using static stimuli in highly controlled experimental settings. Although this provided valuable insights concerning the basic mechanisms underlying visual attention, very little is known about visual attention in dynamic and social situations with less simplified stimuli. Thus, conducting research using closer to real-life situations as stimuli, would significantly contribute to a more complete understanding of visual attention, while it would allow looking into social attention more reliably. This, in turn, could provide a better understanding of phenomena based on social attention such as, social cognition, social learning and the like (click on the picture)

In the Vanaheim project this approach was used to determine a baseline for behavior occurring in specific settings and for control and reliability scoring of computer vision algorithms.