PET: An Eye-tracking Dataset for Animal-centric PASCAL Object Classes

TL;DR

This paper introduces PET, an eye-tracking dataset for animal classes in PASCAL VOC, capturing gaze data in free-viewing and search tasks, and demonstrates its utility for improving animal object classification.

Contribution

The PET dataset is the first to record eye movements for animal classes in both free-viewing and search conditions, providing valuable data for visual attention research.

Findings

Similar fixation counts on target objects across conditions.

Eye-tracking data improves animal object classification accuracy.

Differences in gaze behavior between free-viewing and search tasks.

Abstract

We present the Pascal animal classes Eye Tracking database. Our database comprises eye movement recordings compiled from forty users for the bird, cat, cow, dog, horse and sheep {trainval} sets from the VOC 2012 image set. Different from recent eye-tracking databases such as \cite{kiwon_cvpr13_gaze,PapadopoulosCKF14}, a salient aspect of PET is that it contains eye movements recorded for both the free-viewing and visual search task conditions. While some differences in terms of overall gaze behavior and scanning patterns are observed between the two conditions, a very similar number of fixations are observed on target objects for both conditions. As a utility application, we show how feature pooling around fixated locations enables enhanced (animal) object classification accuracy.