Benchmark RadarGBU-Palm
GBU-Palm is a large-scale multimodal video dataset for palm presentation attack detection, containing 21,326 videos from 105 subjects across six acquisition environments, including bona fide, Print, and Replay attacks, with 6,310 synchronized RGB-NIR samples. It evaluates video architectures under environment-matched and held-out-environment protocols, using metrics such as true accept, true reject, false accept, and false reject rates.
- Released
- 2026-08-14
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Existing palm PAD datasets are limited by static imagery and restricted conditions, hindering systematic evaluation. GBU-Palm provides a unified benchmark to assess robustness across environments and modalities, helping practitioners choose architectures that generalize under environmental shift.
Motivation
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.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.