Saliency and optical flow for gaze guidance in videos

Abstract

Computer-based gaze guidance techniques have important applications in computer graphics, data visualization, image analysis, and training. Bailey et al. [2009] showed that it is possible to influence exactly where attention is allocated using a technique called Subtle Gaze Direction (SGD). The SGD approach combines eye tracking with brief image-space modulations in the peripheral regions of the field of view to guide viewer gaze about a scene. A fast eye-tracker is used to monitor gaze in real-time and the modulations are terminated before they can be scrutinized by the viewer's high acuity foveal vision. The SGD technique has been shown to improve spatial learning, visual search task performance, and problem solving in static digital imagery [Sridharan et al. 2012]. However, guiding attention in videos is challenging due to competing motion cues in the visual stimuli. We propose a novel method that uses scene saliency (spatial information) and optical flow (temporal information) to enable gaze guidance in dynamic scenes. The results of a user study show that the accuracy of responses to questions related to target regions in videos was higher among subjects who were gaze guided with our approach compared to a control group that was not actively guided.