A CLIP-based face anti-spoofing framework that jointly handles classifier weight and bias alignment with group-wise loss scaling and orthogonal feature decomposition reports state-of-the-art HTER and AUC on standard benchmarks.
Invariance principle meets in- formation bottleneck for out-of-distribution generalization
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Group-wise Scaling and Orthogonal Decomposition for Domain-Invariant Feature Extraction in Face Anti-Spoofing
A CLIP-based face anti-spoofing framework that jointly handles classifier weight and bias alignment with group-wise loss scaling and orthogonal feature decomposition reports state-of-the-art HTER and AUC on standard benchmarks.