Why we need: Tracking/predicting the Focus

Visual attention usually involves either overt or covert shifts of attention. In an overt attentional shift a rapid, saccadic, eye movement brings the highest resolution region of the retina, the fovea, onto a region of interest. In a covert shift of attention, however, the focus of attention deviates from the locus of fixation and a more sophisticated method, other than simply tracking the eye position, is required to reveal the attended region. The objective of this paper is to develop a technique for inferring the visual task currently being executed by a person based on measurements of the person’s eye movements and to apply task information to tracking and/or predicting the allocation of covert attention resources. The method we are proposing is to use Hidden Markov Models (HMMs) to create a probabilistic framework to infer the viewer’s task from eye movements and to use this information to select the corresponding task-specific HMM. The Tasks-specific HMM, then, can be used to track and/or predict the allocation of covert attention. What we have: Yarbus process Eye movements measured by Yarbus [1] by viewers carrying out different tasks. Upper right- no specific task, lower left- estimate the wealth of the family, lower right-give the ages of the people in the painting