Training a Camera to Perform Long-Distance Eye Tracking by Another Eye-Tracker

Wenyu Li,Hao Jia,Tianming Liu,Qiang Pan,Li Xie,Qinglin Dong,Feng Duan,Shijie Zhao,Yongchen Wang

doi:10.1109/access.2019.2949150

Abstract

Appearance-based gaze estimation techniques have been greatly advanced in these years. However, using a single camera for appearance-based gaze estimation has been limited to short distance in previous studies. In addition, labeling of training samples has been a time-consuming and unfriendly step in previous appearance-based gaze estimation studies. To bridge these significant gaps, this paper presents a new long-distance gaze estimation paradigm: train a camera to perform eye tracking by another eye tracker, named Learning-based Single Camera eye tracker (LSC eye-tracker). In the training stage, the LSC eye-tracker simultaneously acquired gaze data by a commercial trainer eye tracker and face appearance images by a long-distance trainee camera, based on which deep convolutional neural network (CNN) models are utilized to learn the mapping from appearance images to gazes. In the application stage, the LSC eye-tracker works alone to predict gazes based on the acquired appearance images by the single camera and the trained CNN models. Our experimental results show that the LSC eye-tracker enables both population-based eye tracking and personalized eye tracking with promising accuracy and performance.

Highlights

Gaze estimation has been studied over a few decades and it continues to remain as an interesting research topic [1], [2]
To overcome the above-mentioned key limitations, this paper presents a novel long-distance gaze estimation paradigm: train a single camera to perform eye tracking by another eye tracker, named Learning-based Single Camera eye tracker (LSC eye-tracker)
We proposed a novel longdistance gaze estimation paradigm called LSC eye-tracker, which used a single trainee camera to acquire appearance images from long distance and a trainer commercial eye tracker for gaze data collection simultaneously

Summary

Introduction

Gaze estimation has been studied over a few decades and it continues to remain as an interesting research topic [1], [2]. Application domains of gaze estimation include medical diagnoses and analysis [3], human computer interaction [4], [5], psychological research [6], [7], computer vision [1], product design [8], among many other areas [1], [8]. The equipment and methodology for eye tracking have evolved for several generations [2]. The earliest generation of eye moving measurement consists of scleral contact lens, search coil, EOG and etc. In the second generation of eye tracking equipment, researchers developed photo-oculography and video-oculography for gaze estimation [2]. In the third and fourth generations, The associate editor coordinating the review of this manuscript and approving it for publication was Shovan Barma

Objectives

Results

Conclusion