Reconciling saliency and object center-bias hypotheses in explaining free-viewing fixations

TL;DR

This paper investigates the roles of low-level saliency and high-level object center-bias in guiding visual fixations, demonstrating that a combined model of both cues outperforms individual models and reconciles previous hypotheses.

Contribution

It confirms the importance of object center-bias in fixations and introduces a simple combined model that significantly improves prediction accuracy over existing models.

Findings

Combined model outperforms individual saliency or object bias models

Both low-level saliency and object center-bias are crucial in guiding fixations

Model performs well on diverse scene types and datasets

Abstract

Predicting where people look in natural scenes has attracted a lot of interest in computer vision and computational neuroscience over the past two decades. Two seemingly contrasting categories of cues have been proposed to influence where people look: \textit{low-level image saliency} and \textit{high-level semantic information}. Our first contribution is to take a detailed look at these cues to confirm the hypothesis proposed by Henderson~\cite{henderson1993eye} and Nuthmann \& Henderson~\cite{nuthmann2010object} that observers tend to look at the center of objects. We analyzed fixation data for scene free-viewing over 17 observers on 60 fully annotated images with various types of objects. Images contained different types of scenes, such as natural scenes, line drawings, and 3D rendered scenes. Our second contribution is to propose a simple combined model of low-level saliency and object center-bias that outperforms each individual component significantly over our data, as well as on the OSIE dataset by Xu et al.~\cite{xu2014predicting}. The results reconcile saliency with object center-bias hypotheses and highlight that both types of cues are important in guiding fixations. Our work opens new directions to understand strategies that humans use in observing scenes and objects, and demonstrates the construction of combined models of low-level saliency and high-level object-based information.