University of Maryland scientists used an AI-driven method to recreate 3D scenes based on reflections in the human eye.

University of Maryland scientists used an AI-driven method to recreate 3D scenes based on reflections in the human eye.  The research builds upon Neural Radiance Fields (NeRF), an AI technology that generates 3D representations based on multiple 2D images. By capturing high-resolution images of individuals in motion and examining the reflections in their eyes, the researchers were able to identify objects in the surrounding environment. To remove iris artifacts from the eye images, the researchers relied on texture decomposition. They trained a 2D texture map to learn and eliminate the iris texture present in the images. In addition, cornea geometry was used to determine eye movements and camera angles for accurate scene reconstruction.