The aim of the study was to build a low-cost mask-type eye tracker with accuracy and precision levels similar to those reported for commercial eye tracking devices. To this end, head-mounted hardware was designed and developed, while open-source software was modified for digital image capture, manipulation, and fixation analysis. An image recognition application was also included with different lighting scenarios. Moreover, parallax and viewing perspective errors were controlled to ensure the quality of data collection. The device was wireless and lightweight (99 g) to allow for natural movement and avoid participant discomfort. After calibration of a 9-target monocular grid, spatial accuracy and precision of the eye tracker was evaluated by 30 participants, at four different lighting setups, both before and after a climbing task. Validity tests showed high levels of accuracy in all conditions as evidenced by a systematic error for a 13-target grid of <0.5°. The reliability tests also showed consistent measurements with no differences in accuracy recorded between participants, lighting conditions, and visual behaviors for the pre- versus post-climbing task. These results suggest that the present eye tracker reports spatial accuracy similar to other commercial systems with levels of high quality. Altogether, this innovative user interface is suitable for research purposes and/or performance analysis in physical activity and sport-related activities. Also, features of this mask-type eye tracking system make it a suitable perceptual user interface to investigate human–computer interactions in a large number of other research fields including psychology, education, marketing, transportation, and medicine.
|Journal||Proceedings of the Institution of Mechanical Engineers, Part P: Journal of Sports Engineering and Technology|
|Early online date||9 Nov 2018|
|Publication status||E-pub ahead of print - 9 Nov 2018|