Advancing Cross-Domain Generalizability in Face Anti-Spoofing: Insights, Design, and Metrics

arXiv:2406.12258v1 Announce Type: new
Abstract: This paper presents a novel perspective for enhancing anti-spoofing performance in zero-shot data domain generalization. Unlike traditional image classification tasks, face anti-spoofing datasets display unique generalization characteristics, necessitating novel zero-shot data domain generalization. One step forward to the previous frame-wise spoofing prediction, we introduce a nuanced metric calculation that aggregates frame-level probabilities for a video-wise prediction, to tackle the gap between the reported frame-wise accuracy and instability in real-world use-case. This approach enables the quantification …

abstract arxiv classification cs.cv data datasets design display domain face image insights metrics novel paper performance perspective prediction tasks type unique wise zero-shot