Demonstrating Task-Aware Saliency for 3D Shapes

Anonymous

Keywords: mesh saliency, 3D graphics

TL;DR: While the concept of task-aware saliency has been demonstrated for 2D images, we give evidence in this paper that this concept is also true for 3D shapes.

Abstract: The concept of task-aware saliency has been demonstrated for 2D images. The idea of this concept is that for different tasks, where people look or focus their attention (i.e. saliency) on 2D images is different. However, this concept has not been demonstrated for 3D shapes. In this paper, we collect data on where people look on 3D shapes while performing different tasks (e.g. classifying a shape, deciding how aesthetic a shape is, or describing a shape). We then display the data as task-aware saliency maps on the 3D shapes, and demonstrate that where people look on 3D shapes as they are performing different tasks is different.