Edge-Assisted Object Detection Using Eye Tracking Function of Mixed Reality Devices

Although object detection methods have significant potential to extend the boundaries of mixed reality (MR), they are often difficult to execute on current MR devices that have limited computing resources. In this paper, we propose an edge-assisted object detection method that uses the eye-tracking function of MR devices to determine whether an object detection task should be offloaded to an edge server for processing and thus reduce the MR device processing load. We implement and evaluate the proposed method using a Microsoft HoloLens 2 to confirm its effectiveness.