3D Gaze Recovery in Large Environments Using Visual SLAM

Publication from Digital

Lucas Paletta, Katrin Santner and Albert Hofmann and Georg Thallinger

Proceedings of the First International Workshop on Solutions for Automatic Gaze-Data Analysis 2013 (SAGA 2013) , 1/2013


This work describes a multi-component vision system that enables pervasive map-ping of human attention. The key contribution is that our methodology enables full 3D recovery of the gaze pointer, human view frustum and associated human cen-tered measurements directly into an automatically computed 3D model. We apply RGB-D SLAM and descriptor matching methodologies for the 3D modeling, lo-calization and fully automated annotation of ROIs (regions of interest) within the acquired 3D model. This methodology enables fully automated processing of hu-man attention, without artificial landmarks, in indoor natural environments.