Vision Transformers reduce demographic bias in face anti-spoofing systems

By PulseAugur Editorial · [1 sources] · 2026-06-16 22:00

A new study published on arXiv investigates the impact of Vision Transformer (ViT) architectures on demographic bias in face presentation attack detection (PAD) systems. The research compares ViTs against convolutional neural networks (CNNs) using the CASIA-SURF Cross-Ethnicity Face Anti-Spoofing (CeFA) dataset. Results indicate that ViT models, specifically the DeiT-S architecture, achieve higher accuracy and significantly reduce performance disparities across different ethnic groups compared to CNN baselines. AI

IMPACT Vision Transformer architectures may offer a path toward more equitable and robust face anti-spoofing systems, reducing bias against darker skin tones.

RANK_REASON Academic paper detailing a comparative study of AI model architectures. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

paper
safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Vision Transformers reduce demographic bias in face anti-spoofing systems

COVERAGE [1]

arXiv cs.CV TIER_1 English(EN) · Jema David Ndibwile · 2026-06-16 22:00

Architectural Bias in Face Presentation Attack Detection: A Comparative Study of Vision Transformers and Convolutional Neural Networks

Face Presentation Attack Detection (PAD) systems constitute a critical security layer in biometric authentication; however, existing approaches exhibit systematic performance disparities across demographic groups, disproportionately affecting individuals with darker skin tones. T…

COVERAGE [1]

Architectural Bias in Face Presentation Attack Detection: A Comparative Study of Vision Transformers and Convolutional Neural Networks

RELATED ENTITIES

RELATED TOPICS