GBU-Palm: A Multimodal Video Dataset and Benchmark for Palm Presentation Attack Detection

2026-08-14

AI Summary

This study addresses the scarcity of multimodal video data and robustness evaluation challenges in palmprint liveness detection by constructing the first large-scale video benchmark comprising over 20,000 synchronized RGB-NIR samples. A leakage-prevention protocol decoupling identity from attack lineage is introduced to ensure rigorous assessment. Systematic evaluations reveal significant performance degradation under cross-environment transfer and demonstrate that RGB-NIR fusion does not consistently outperform unimodal approaches, while elucidating distinct failure modes and evidence utilization patterns across architectures. By establishing new standards for cross-environment robustness evaluation, this work provides a unified platform and critical insights for advancing multimodal palmprint liveness detection research.

Abstract

Existing palm presentation attack detection (PAD) datasets are often limited by static imagery, restricted acquisition conditions, or insufficient multimodal video data, hindering systematic evaluation across environments, modalities, and attack types. We present GBU-Palm, a large-scale multimodal video dataset and benchmark containing 21,326 videos from 105 subjects and 210 palms across six acquisition environments, including bona fide, Print, and Replay presentations, with 6,310 synchronized RGB-NIR samples. We construct leakage-controlled protocols that separate palm identity and attack lineage and benchmark four representative video architectures under environment-matched and held-out-environment settings. Results reveal substantial architecture-dependent degradation under environmental shift and show that RGB-NIR fusion does not consistently outperform RGB-only input. We further analyze model behavior through true accept (TA), true reject (TR), false accept (FA), and false reject (FR) decomposition, spectral masking, temporal-order intervention, and frozen-backbone NIR probing, revealing distinct failure patterns and evidence utilization across architectures. GBU-Palm provides a unified and challenging benchmark for developing and evaluating robust multimodal palm PAD methods under cross-environment conditions.

Problem

• Palm Presentation Attack Detection
• Multimodal Video Dataset
• Cross-environment Evaluation
• Benchmark

Innovation

• Multimodal Video Dataset
• Palm Presentation Attack Detection
• Cross-environment Benchmark
• Leakage-controlled Protocol
• RGB-NIR Fusion