Object-based Saliency as a Predictor of Attention in Visual Tasks

Abstract

The top-down guidance of visual attention is an important fac- tor allowing humans to effectively process incoming visual information. Our understanding of the processes governing attention is not complete, with growing evidence for attention selection based on cognitive relevance. In this paper, we investigate whether models for salient object detection from computer vision can be used to predict attentional shifts in visual tasks. Our results show that the object-based interpretation of saliency provided by these models is a substantially bet- ter predictor of fixation locations than traditional pixel-based saliency.


Back to Table of Contents