Get Started
Home
Topics
Search
Library
Research questionHow can gaze trajectories be synthesized for biometric authentication while preserving identity, task context, and subjective state?Gaze recordings are expensive to collect at the scale needed for biometric model development. Synthetic trajectories may look realistic yet fail to preserve the behavioral cues associated with identity, task demands, or mental and eye fatigue.
AI
Computer Vision
Diffusion Models
Machine Learning
Research Paper
Technology
Latest papersRecent research connected to this question, newest first.EyeMakeYou: Identity-, Task-, and Subjective-State-Conditioned Diffusion for High-Frequency Gaze SynthesisThe source studies 5-second, 1000-Hz bivariate gaze-velocity synthesis from identity-removed reference trajectories, conditioned on identity, task, and self-reported difficulty and fatigue ratings. Its evidence comes from GazeBase, where the proposed diffusion framework reports higher median spatial accuracy, greater real–synthetic embedding similarity, and preservation of selected task-dependent associations.research paper · Sep 3, 2026
Related questions
How should gaze representations balance XR task utility against biometric identity leakage?How can online gaze correction stabilize target acquisition by modeling task phase and persistent residuals without target information?How can mask-free video virtual try-on maintain garment consistency under motion, occlusion, and changing viewpoints?How can robots recognize social gestures in real time despite occlusion, viewpoint changes, and noisy keypoints?