GlintMarkers: Gaze-Anchored Spatial Perception Using Corneal Reflections
Abstract
AI assistants on smart glasses need to know what surrounds the user and what the user is looking at. Obtaining this context typically relies on a world-facing camera, raising privacy concerns for bystanders. We present GlintMarkers, a system for gaze-anchored spatial perception using a single inward-facing eye camera. Our key observation is that the cornea acts as a mirror that encodes both gaze direction and visual information about the environment in a small, low-contrast reflection. To extract spatial and semantic information from this reflection despite the camera's limited pixel budget, we design passive retroreflective markers that concentrate reflected near-infrared light onto the cornea, producing bright glint patterns. We develop a custom Perspective-n-Point (PnP) estimation framework adapted to corneal imaging and perform orientation and distance estimation of marked objects, as well as unique object identification.
