Automatic eye landmark detection combines CNN estimation with post-hoc refinement to improve biometric measurement accuracy, speed, and repeatability.
Relative motion of rigid and non-rigid facial features helps detect spoofed faces from short mobile videos without complex hardware.
Gaze, pupil, and hand-position signals are combined into team awareness metrics to assess remote collaboration and flag needed interventions.
Tracks facial key-point changes across image sequences to detect blinks or mouth motion and block prosthesis-based spoofing attacks.
Modulated on-axis light creates a bright pupil signal that frequency segmentation can separate from ambient reflections for more accurate eye tracking.
By identifying the colored lens pattern and excluding its tinted region, iris matching stays accurate even when colored contacts are worn.
Eye cameras capture the user's face during headset on/off motion to build a 3D model for realistic avatars without extra user actions.
Corneal reflection and pupil coordinates are estimated from face images to improve gaze measurement accuracy without added hardware complexity.
Self-mixing interferometry tracks subtle eye motion with lower power and latency than camera-based gaze sensing in head-mounted displays.
Eye-image AI extracts iris and color features to infer ancestral information and flag aberrations that support self-awareness.
Single-shot asperity-based 3D subject modeling enables pose and form changes without repeated retakes, improving image output flexibility.
Eye-feature AI extracts iris, color, and structure cues to infer ancestral information and flag unresolved trauma with staged analysis.