Related Experiment Video
Updated: Aug 1, 2025
09:27
Using Eye Movements Recorded in the Visual World Paradigm to Explore the Online Processing of Spoken Language
Published on: October 13, 2018
10.1K
A field test of computer-vision-based gaze estimation in psychology
Niilo V Valtakari1, Roy S Hessels2, Diederick C Niehorster3,4
1Experimental Psychology, Helmholtz Institute, Utrecht University, Heidelberglaan 1, 3584 CS, Utrecht, the Netherlands. niilovaltakari@gmail.com.
Behavior Research Methods
|April 26, 2023
Summary
OpenGaze is accurate for screen-based experiments, while OpenFace may suit sparser environments. Both are computer-vision gaze estimation methods usable by researchers without eye trackers.
Area of Science:
- Psychology
- Education
- Computer Science
- Human-Computer Interaction
Background:
- Computer-vision-based gaze estimation offers an alternative to traditional eye trackers.
- Validation of these methods is often limited to technical literature, hindering adoption in other fields.
Purpose of the Study:
- Identify and evaluate computer-vision gaze estimation methods suitable for researchers in psychology and education.
- Assess methods requiring no calibration and possessing clear documentation.
Main Methods:
- Two toolkits, OpenFace and OpenGaze, were selected based on criteria.
- Experiments involved adult participants fixating on screen stimuli, with video recordings processed by the toolkits.
- Further analysis explored OpenFace's utility with infant participants and sparse stimuli.
Main Results:
- OpenGaze demonstrated sufficient accuracy and precision for screen-based experiments with stimuli separated by at least 11 degrees.
- OpenFace showed lower accuracy for dense stimuli but potential for sparser environments.
- OpenFace estimates showed promise for relative dwell time measures in sparse environments with infants, but not for absolute dwell duration.
Conclusions:
- OpenGaze is a viable tool for screen-based gaze research in psychology and education.
- OpenFace may be applicable in specific, sparser research contexts, particularly for relative gaze measures.
- Further research is needed to refine the application of computer-vision gaze estimation in diverse research settings.
