Semantic 3D gaze mapping toward comprehending attention of multiple people
Ryusei MATSUMOTO, Kentaro Takemura · The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) · 2019
Gaze information conveys what the people are interested in, and it is expected that the analysis of the point-of-gaze can be used for marketing research. However, the point-of-gaze is computed on a screen, and it is difficult to estimate the 3D point-of-gaze in the world-coordinate system for gaze data of multiple people. Therefore, we propose a method for estimating the 3D point-of-gaze using structure from motion. The semantic 3D map is reconstructed using a semantic segmentation and keyframes selected from scene images. Additionally, the 3D point-of-gaze can be also estimated on the 3D map without prior information.