Gaze Estimation from Multimodal Kinect Data

This demo shows qualitative results for our paper presented at the Gesture Recognition Workshop at CVPR 2012. The main steps of the processing pipeline are shown, including the head pose tracking and the face appearance stabilization technique, from which we extract the frontal looking eye images. For details please refer to the paper.

Geometric Generative Gaze Estimation (G3E) for Remote RGB-D Cameras

The following is the video spotlight for our CVPR 2014 paper, in which we briefly describe the motivation for Geometric Generative Gaze Estimation (G3E) model. For details please refer to the paper.

Gaze Coding in Natural Dyadic Interactions

The following video shows results for the problem of automatic gaze coding in natural dyadic interactions. The gaze model used in both subjects is person invariant, meaning that there was no subject-specific training being done. These results are novel, but the closest methodology can be found in this paper.