Object-wise 3D Gaze Mapping in Physical Workspace

Title: Object-wise 3D Gaze Mapping in Physical Workspace

Teams: Keio University

Writers: Kakeru Hagihara;Keiichiro Taniguchi;Irshad Abibouraguimane;Yuta Itoh;Keita Higuchi;Jiu Otsuka;Maki Sugimoto;Yoichi Sato

Publication date: February 2018

Abstract

Understanding the intention of other people is a fundamental social skill in human communication. Eye behavior is an important, yet implicit communication cue. In this work, we focus on enabling people to see the users’ gaze associated with objects in the 3D space, namely, we present users the history of gaze linked to real 3D objects. Our 3D gaze visualization system automatically segments objects in the workspace and projects user’s gaze trajectory onto the objects in 3D for visualizing user’s intention. By combining automated object segmentation and head tracking via the first-person video from a wearable eye tracker, our system can visualize user’s gaze behavior more intuitively and efficiently compared to 2D based methods and 3D methods with manual annotation. We performed an evaluation of the system to measure the accuracy of object-wise gaze mapping. In the evaluation, the system achieved 94% accuracy of gaze mapping onto 40, 30, 20, 10-centimeter cubes. We also conducted a case study of through a case study where the user looks at food products, we showed that our system was able to predict products that the user is interested in.