Abstract
TASK RELEVANCE VS SALIENCY IN COMPUTATIONAL MODELLING OF DIRECTED ATTENTION FATIGUE
James Mountstephens* and Toh Chia Ming
ABSTRACT
Failures of visual attention can be life threatening. Computer models of visual attention have not previously addressed the issue of fatigue over time and also its underlying mechanisms. This work explores the hypothesis that directed attention fatigue is caused by an inability to inhibit responses to salient distractors in favour of task-relevant targets. A model of this competition in gaze allocation was constructed using the Matlab Saliency toolbox implementation of Itti and Koch?s bottom-up attention model which was extended with foveation, decision making, and a task-relevance map. Human performance and gaze data was collected by eye tracker during a novel mentally-fatiguing task and used to tune and evaluate the model. The initial results are encouraging.
[Full Text Article] [Download Certificate]