IriTrack: Face Presentation Attack Detection Using Iris Tracking
summary
IriTrack challenges users to follow a randomly generated 3-dot poly-line with their eyes, records iris centers using Daugman’s integro-differential operator, and flags attacks when trajectory angles deviate past a threshold. A prototype runs on a commercial PC, tested with 4 participants × 10 trials per pattern, compares against three PAD baselines using APCER and BPCER, and reports the highest detection accuracy with moderate time/memory cost while remaining robust to lighting, distance, and head-pose variation.
AI-written summary · about
related talks
14mBreaking Infrared Recapture Detection: Optical-Synthesis Attacks and Depth-Aware...
14mRed Bleed: A Pragmatic Near-Infrared Presentation Attack on Facial Biometric...
26mVirtual U: Defeating Face Liveness Detection by Building Virtual Models...
26mNew Face, Who Dis? Recent Adversarial Approaches to Facial Recognition
28mFace spoofing detection: theory and practice
45mYour passport to the 7th continent
32mHeart to Heart and Eye to Eye with Deep Fakes
17mReflector: Distance-Independent, Private Pointing on a Reflective Screen