The field of human-computer interaction is moving towards more accurate and efficient methods for gaze estimation and human localization. Recent developments have focused on leveraging egocentric cues, such as gaze direction and head-mounted IMU signals, to improve the accuracy of gaze estimation and inertial localization. These advancements have the potential to enable seamless interactions in applications such as smart glasses and human-robot interaction. Notably, novel architectures for gaze estimation have been proposed, which can predict 6-DoF gaze information and achieve state-of-the-art performance. Furthermore, innovative approaches for inertial localization have been developed, which can effectively compensate for localization drift and recognize sequences of actions.

Noteworthy papers include: GA3CE, which proposes a novel 3D gaze estimation approach that learns spatial relationships between the subject and objects in the scene. MAGE, which introduces a multi-task architecture for gaze estimation with an efficient calibration module to predict 6-DoF gaze information. Egocentric Action-aware Inertial Localization, which presents a novel inertial localization framework that leverages egocentric action cues from head-mounted IMU signals to localize the target individual within a 3D point cloud.