GPU-accelerated attention map generation for dynamic 3D scenes

Abstract

Measuring visual attention has become an important tool during product development. Attention maps are important qualitative visualizations to communicate results within the team and to stakeholders. We have developed a GPU-accelerated approach which allows for real-time generation of attention maps for 3D models that can, e.g., be used for on-the-fly visualizations of visual attention distributions and for the generation of heat-map textures for offline high-quality renderings. The presented approach is unique in that it works with monocular and binocular data, respects the depth of focus, can handle moving objects and is ready to be used for selective rendering.