December 2022
Volume 22, Issue 14
Open Access
Vision Sciences Society Annual Meeting Abstract  |   December 2022
Evaluating Data Stability During Active Head-Eye Tracking: A Comparison of Dynamic Gaze Error between Two Custom-Built Head-Mounted Devices
Author Affiliations & Notes
  • Kamran Binaee
    University of Nevada Reno
  • Bharath Shankar
    University of Nevada Reno
  • Brian Szekely
    University of Nevada Reno
  • Michelle Greene
    Bates College
  • Paul MacNeilage
    University of Nevada Reno
  • Footnotes
    Acknowledgements  Research was supported by NSF Award number 1920896: RII Track-2 FEC: The Visual Experience Database: A Large-Scale Point-of-View Video Database for Vision Research by Dr. Michelle R. Greene
Journal of Vision December 2022, Vol.22, 4469. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Kamran Binaee, Bharath Shankar, Brian Szekely, Michelle Greene, Paul MacNeilage; Evaluating Data Stability During Active Head-Eye Tracking: A Comparison of Dynamic Gaze Error between Two Custom-Built Head-Mounted Devices. Journal of Vision 2022;22(14):4469.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

The Visual Experience Database is a large-scale effort to capture the high-quality head and gaze data across a wide range of diverse participants and activities. One key goal is to characterize how humans use eye and head movements to acquire information from the environment. However, for active participants in real-life conditions, this requires robust gaze tracking and precise characterization of different sources of error. Slippage, the physical shift of the device on the head, is a major source of error and can limit the types of activities feasible for data collection. In this study, we first introduce two custom-assembled head-eye tracking systems (rigid and elastic) equipped with a heavy high-quality world camera. Next, we propose a novel method to precisely characterize the effect of slippage for each rig throughout the data collection session. We recorded data from 10 participants fixating a stationary target while walking on a treadmill for 90 seconds at three different speeds (1.8, 3.1, and 4.1 mph) corresponding to slow, normal, and fast walking. We repeated this using both devices and ascending or descending speed orders. We use two gaze error metrics: drift, defined as the cumulative error over time, and jitter, defined as the standard deviation of zero-mean error. Average gaze drift in elevation was found to be 0.61 and 1.01 degrees/minute (p<0.02) and instantaneous jitter was 5.01 and 3.99 degrees (p<0.3) for the elastic and rigid devices, respectively. Our findings reveal that the elastic rig provides a smaller drift and a non-significant larger jitter value, proving to be a better choice. Since we have recorded the unperturbed gaze vector i.e. fixation target position during each walking condition, we also discuss the analysis methods and computational models suitable for compensating for both drift and jitter in time.


This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.