A Dataset for Table-Meeting Analysis Using Spherical Camera
Nurul Hidayah Rahim, Ryusuke Nakayama, Shigang Li · 2021 IEEE 10th Global Conference on Consumer Electronics (GCCE) · 2021
A gaze estimation is very important for multi-person discussion such as a table meeting. In this paper, we use a spherical camera to observe multi-person sitting around the table and want to estimate their gaze point. In comparison with the conventional method which uses a camera for each person, we can observe all people simultaneously by using a single spherical camera. Based on this idea, we built a dataset using a spherical camera for four person table-meeting in three different cases, i.e., two people look at a different target, two people look at the same target and three people look at the same target. As the future work, we will use this dataset to train artificial neural network to estimate gaze point and evaluate their accuracy.