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

Yingjie Ma; Zitong Yu; Wei Jia; Ajay Kumar; Linlin Shen

cs.CVAug 14, 2026

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

Authors: Yingjie Ma, Zitong Yu, Wei Jia, Ajay Kumar, Linlin Shen

Organizations: Shenzhen University · Great Bay University · Dongguan Key Laboratory for Intelligence and Information Technology · Guangdong Provincial Key Laboratory of Intelligent Information Processing, Shenzhen University · Hefei University of Technology · The Hong Kong Polytechnic University

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.

Figures & tables

Dataset	Access	Unit	Scale	Modality	Attack
VERA [ 9 ]	Request	Image	2,000	NIR	Print
PALMspoof [ 2 ]	Private	Image	–	RGB	Print/display
XJTU-PalmReplay [ 13 ]	Request	Image	96,000	RGB	Replay
PVASD [ 11 ]	Public	Image	1,187,519	NIR	2D/3D
GBU-Palm	Public	Video	21,326	RGB/NIR	Print/replay

Table 1: Representative palm PAD resources. Scale follows the evaluation unit reported by each source and is not directly comparable across rows.

Attack	Factor	Values
Print	Spectrum	RGB color, grayscale, NIR
Print	Material	A4, photo, matte, pearl, coated paper
Print	Crop	Full, local, hand-shape, partial real-palm exposure
Replay	Spectrum	RGB, NIR
Replay	Interface	Full-screen, UI borders

Table 2: Presentation-attack factors retained in GBU-Palm metadata.

Figure 1: Representative GBU-Palm samples across six acquisition environments (E1–E6) and three presentation classes. RGB and NIR rows show synchronized observations of the same physical presentations, illustrating variation across environment, attack type, and sensing spectrum.

Protocol	Split	Total	Bona fide	Print	Replay	RGB+NIR
P1	Train	12,152	7,433	2,049	2,670	3,740
	Val	3,920	1,649	1,294	977	1,115
	Test	3,422	1,641	902	879	1,031
P2	Train	4,827	3,455	445	927	1,482
	Val	1,975	1,040	618	317	529
	Test	2,174	1,471	279	424	836

Table 3: Split composition of the GBU-Palm P1 (In-Env) and P2 (Cross-Env) protocols. RGB+NIR denotes samples with synchronized paired observations.

Method	P1: In-Env		P2: Cross-Env
Method	AUC $\uparrow$ (%)	HTER $\downarrow$ (%)	AUC $\uparrow$ (%)	HTER $\downarrow$ (%)
R(2+1)D-18 [ 10 ]	97.48	8.39	94.61	13.05
ViViT [ 1 ]	93.49	14.55	82.21	25.73
Video Swin-T [ 8 ]	72.95	32.89	71.20	36.36
MViT-V2-S [ 4 ]	97.33	8.39	93.30	17.00

Table 4: Overall video-based PAD results on the complete GBU-Palm test splits. AUC and HTER are reported in percentage; HTER uses the operating threshold fixed on validation.

Method	P1 Best	P1 Worst	Range	P2 E1	P2 E3
R(2+1)D-18	E1 / 98.7	E6 / 96.5	2.2	95.2	94.1
ViViT	E2 / 95.8	E5 / 91.2	4.6	82.6	81.8
Video Swin-T	E6 / 77.1	E4 / 67.1	10.0	71.7	70.6
MViT-V2-S	E1 / 98.3	E6 / 96.7	1.7	94.6	92.0

Table 5: Environment-specific PAD performance (AUC, %). P1 Range denotes the maximum–minimum AUC across the six P1 environments. P2 values compare the two held-out environments within the same protocol.

Figure 2: Decision-conditioned examples from GBU-Palm. TA, TR, FA, and FR cases show representative temporal observations ( $t_{1}$ – $t_{4}$ ) and corresponding attack-probability trajectories. The visualization highlights different decision behaviors without making unsupported pixel-level attribution claims.

Method	P1: In-Env			P2: Cross-Env
Method	RGB	NIR	RGB+NIR	RGB	NIR	RGB+NIR
R(2+1)D	99.45/4.33	84.09/25.03	99.66/3.02	96.94/7.97	91.54/16.19	98.65/4.08
ViViT	95.61/9.13	67.71/44.54	98.82/4.82	83.75/21.43	67.14/45.41	87.15/19.14
Swin-T	87.42/19.74	30.69/65.64	83.67/22.98	78.84/25.33	77.97/20.52	82.39/21.92
MViTv2-S	97.25/6.66	99.66/2.61	99.74/1.77	97.20/6.77	99.07/5.47	99.38/5.58

Table 6: Controlled spectral and temporal evidence. (a) Paired-test AUC/HTER (%) from the same multimodal-trained checkpoint and identical sample identities. (b) Full-test AUC under normal, shuffled, and reversed temporal order; shuffled results report mean

\pm

std over five fixed permutations. All temporal interventions preserve the same set of observed frames.

Figure 3: Frozen-backbone NIR diagnosis. The shared head uses the original multimodal classifier, while the linear probe is trained on paired-validation NIR embeddings and evaluated on paired test samples.

Explore similar work

X-Palm: Paired Multispectral-to-Smartphone Dataset for Cross-Domain Palmprint Authentication
Jun 7, 2026Jamal Seyedmohammadi, Pai Chet Ng, Angelo Genovese +3Biometric Identification Cross-Domain Generalization
Singapore Institute of Technology · Università degli Studi di Milano · University of Toronto
CAAP: Capture-Aware Adversarial Patch Attacks on Palmprint Recognition Models
Apr 8, 2026Renyang Liu, Jiale Li, Jie Zhang +6Adversarial Patch Attacks Physical Adversarial Attacks
Institute of Data Science, National University of Singapore, Singapore 117602, Singapore · Centre for Frontier AI Research (CFAR), A*STAR, Singapore, 138634 · School of Cyber Science and Engineering, Wuhan University, China, 430072 +2
From Vision to Text: A Compact Multimodal Approach for Robust, Cross-Domain Presentation Attack Detection on ID Cards
Jun 5, 2026Qingwen Zeng, Juan E. Tapia, Sneha Das +1Multimodal Robustness Domain Generalization
Technical University of Denmark (DTU), Denmark · da/sec-Biometrics and Security Research Group, Hochschule Darmstadt, Germany

Jun 7, 2026J/K move · Enter open · S save

X-Palm: Paired Multispectral-to-Smartphone Dataset for Cross-Domain Palmprint Authentication

Jamal Seyedmohammadi, Pai Chet Ng, Angelo Genovese, Zhixiang Chi+2

Singapore Institute of Technology · Università degli Studi di Milano · University of Toronto

Biometric Identification Cross-Domain Generalization

Open paper AskPDF arXiv

Palmprint modality offers a privacy-preserving biometric solution, yet its deployment is hindered by the domain gap between controlled enrollment and unconstrained authentication. Existing datasets are largely restricted to controlled setups and fail to capture the compound variability of real-world environments. In this paper, we introduce X-Palm, a cross-domain dataset comprising 6,006 palm images from 103 individuals (206 hands). To the best of our knowledge, X-Palm is the first palmprint dataset providing novel paired-identity acquisition specifically designed to bridge the gap between reliably controlled multispectral enrollment and unconstrained mobile authentication while encompassing a broad spectrum of in-the-wild variability. Unlike existing datasets that focus on single to a few variations, X-Palm addresses the massive modality and environmental shifts encountered in practical deployments by capturing paired data for identities across two distinct domains: (1) a controlled Multispectral Palmprint setting using our custom-developed scanner, and (2) an unconstrained smartphone palmprint setting that is participant-driven, incorporating simultaneous variations in hardware, hand pose, illumination, background, camera-to-hand distance, perspective, and palm surface conditions (e.g., moisture and occlusions). Our extensive benchmarks of 12 SOTA models reveal that while existing methods achieve high performance on controlled data, they experience severe performance collapse on X-Palm. Conversely, models trained on X-Palm demonstrate consistent robustness across domains, positioning X-Palm as a valuable resource for training a model towards real-world, cross-domain generalization. Data access instructions and the related benchmarking codes are publicly available at: https://github.com/X-Palm/X-Palm-2026

Abstract

Explore similar work

X-Palm: Paired Multispectral-to-Smartphone Dataset for Cross-Domain Palmprint Authentication

CAAP: Capture-Aware Adversarial Patch Attacks on Palmprint Recognition Models

From Vision to Text: A Compact Multimodal Approach for Robust, Cross-Domain Presentation Attack Detection on ID Cards