Eye behavior has gained much interest in the VR research com- munity as an interactive input and support for collaboration. Re- searchers used head behavior and saliency to implement gaze in- ference models when eye-tracking is missing. However, these solu- tions are resource-demanding and thus unfit for untethered devices, and their angle accuracy is around 7°, which can be a problem in high-density informative areas. To address this issue, we propose a lightweight deep learning model that generates the probability density function of the gaze as a percentile contour. This solution allows us to introduce a visual attention representation based on a region rather than a point. In this way, we manage the trade-off between the ambiguity of a region and the error of a point. We tested our model in untethered devices with real-time performances; we evaluated its accuracy, outperforming our identified baselines (average fixation map and head direction).

Riccardo Bovo, Daniele Giunchi, Ludwig Sidenmark, Enrico Costanza, Hans Gellersen, and Thomas Heinis. 2022. Real-time head-based deep-learning model for gaze probability regions in collaborative VR. In ETRA ’22: ACM Symposium on Eye Tracking Research and Applications, June 08–11, 2022, Seattle, WA. ACM, New York, NY, USA, 8 pages.