Gaze estimation problem tackled through synthetic images

Abstract

In this paper, we evaluate a synthetic framework to be used in the field of gaze estimation employing deep learning techniques. The lack of sufficient annotated data could be overcome by the utilization of a synthetic evaluation framework as far as it resembles the behavior of a real scenario. In this work, we use U2Eyes synthetic environment employing I2Head datataset as real benchmark for comparison based on alternative training and testing strategies. The results obtained show comparable average behavior between both frameworks although significantly more robust and stable performance is retrieved by the synthetic images. Additionally, the potential of synthetically pretrained models in order to be applied in user's specific calibration strategies is shown with outstanding performances.

Low-Cost Eye Tracking Calibration: A Knowledge-Based Study

“…It is here where synthetic databases, which provide absolute control of the ground-truth, can be an essential support point for the development of research in this field. In the last few years, works from different areas have benefited from training over synthetic environments [ 6 , 7 , 8 , 9 ].…”