3D head pose and gaze tracking using remote RGB-D sensors
Kenneth A. Funes Mora, EPFL
Gaze is acknowledged as one of the most important non-verbal cues in human communication. However, its automatic estimation is a highly challenging problem, in particular, when large user mobility is expected and minimal intrusion is required. During the past years, we have investigated techniques to remotely sense the gaze direction using consumer RGB-D (color and depth) cameras. In this talk, I will discuss the main challenges associated to this task: low resolution, head pose variation, inter-user appearance variation, etc.; and how we have addressed them. As an example application, I will discuss the automatic gaze coding of natural dyadic interactions. Overall, we aim to develop a remote 3D gaze sensing system which can represent an alternative to the highly costly and/or intrusive systems currently available. Such a solution would have an important potential in the development of diverse human-human, human-robot and human-computer interaction applications and studies.