Pith. sign in

REVIEW 4 major objections 4 minor 54 references

Exploring Task-Solving Paradigm for Generalized Cross-Domain Face Anti-Spoofing via Reinforcement Fine-Tuning

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Reinforcement fine-tuning with binary real/spoof labels alone lets a multimodal vision-language model generalize across face-spoofing domains and beat supervised baselines on four cross-domain protocols.

desk verdict A genuinely new application of RL fine-tuning to face anti-spoofing, with a promising idea and a readable write-up, but the headline SOTA numbers rest on an underspecified baseline comparison that needs to be pinned down before the claims can be trusted. read the letter →

arxiv 2506.21895 v1 pith:LVETOGHG submitted 2025-06-27 cs.CV

Fangling Jiang , Qi Li , Weining Wang , Gang Wang , Bing Liu , Zhenan Sun This is my paper · ORCID
classification cs.CV
keywords faceanti-spoofingcross-domaingeneralizationreinforcementfine-tuninggrouprelativepolicyoptimizationmultimodallargelanguagemodelpresentationattackdetectioninterpretablereasoningrewarddesign
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that a multimodal large language model can learn to solve face anti-spoofing as a task, rather than memorize training patterns, by fine-tuning it with reinforcement learning on binary real/spoof labels alone. It reports that the resulting model, Qwen2.5-VL-7B-Instruct fine-tuned with GRPO and three verifiable rewards, beats prior supervised methods on four cross-domain protocols, with relative HTER reductions of 18.48%, 60.75%, 18.17%, and 28.28% over the second-best method. If this holds, anti-spoofing systems could be built without labor-intensive textual reasoning annotations, and they would generalize to unseen attack types such as makeup, masks, and mannequins while giving interpretable explanations for each decision. The paper's contribution is therefore a learning paradigm shift, reward-guided exploration of decision policies, rather than a new spoof-feature descriptor.

What carries the argument

The load-bearing mechanism is Group Relative Policy Optimization (GRPO) applied to a multimodal policy model, with three verifiable rewards: a format reward that enforces the <think>/<answer> structure, a class-consistent reward that gives 1 when the predicted class matches the ground truth and 0 otherwise, and a reasoning-consistent reward that scales positively with reasoning length for correct predictions and negatively for incorrect ones. GRPO samples N responses per image, normalizes each response's total reward by the group mean and standard deviation to get an advantage, and updates the policy with a clipped objective plus a KL-divergence constraint to the reference model. This reward structure is what forces the model to explore many reasoning policies and retain only those that classify correctly, distilling generalizable decision rules rather than memorized answer formats.

What would settle it

Re-run the four cross-domain protocols with every baseline scored by the same direct text-prediction rule, or with thresholds set identically on the source domain only, and check whether the relative HTER gains of 18.48%, 60.75%, 18.17%, and 28.28% persist; if the gaps shrink or reverse, the state-of-the-art claim is falsified.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that reinforcement fine-tuning with only real/fake labels lets a vision-language model discover transferable decision rules for face anti-spoofing. The model is prompted to reason in <think>...</think> tags and answer real or spoof; its responses are scored by format, class consistency, and reasoning-consistency rewards, and GRPO keeps only high-reward trajectories while discouraging drift from the reference policy. Across CeFa to HQ-WMCA, CeFa to SiW-Mv2, CASIA-SURF to HQ-WMCA, and CASIA-SURF to SiW-Mv2, the method reports HTER values of 15.75%, 8.89%, 16.48%, and 9.74%, which the paper states are relative improvements of 18.48%, 60.75%, 18.17%, and 28.28% over the second-best method, using direct text-prediction accuracy without any threshold. The same model also handles unseen attacks such as makeup, transparent masks, paper glasses, and mannequins, and its visible reasoning lines up with standard anti-spoofing cues like texture unnaturalness, lighting, distortion, and 3-D appearance.

Load-bearing premise

The load-bearing premise is that the baselines were scored under the same decision rule and threshold logic as the proposed method; the paper reports no details on how baseline thresholds were set, while the proposed method uses direct text-prediction without any threshold.

Editorial extensions

If this is right

  • A face anti-spoofing model can be trained from binary real/spoof labels alone, removing the need for manually written reasoning annotations.
  • The same fine-tuning recipe should transfer to other unseen attack families: the paper shows low error on makeup, masks, paper glasses, mannequins, and wigs.
  • Decisions come with readable chain-of-thought explanations that refer to texture, lighting, distortion, and 3-D appearance, which is what deployment and auditing would want.
  • More training data and larger policy sampling group sizes both improve HTER, suggesting the method scales with exploration rather than with annotation effort.
  • Evaluation can be done by comparing direct text predictions to labels, avoiding threshold selection.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: if the reported gains survive a matched decision-rule comparison, reward design rather than feature engineering would become the main lever in face anti-spoofing research.
  • Editorial inference: the reasoning-consistent reward's length-based penalty for wrong answers could create a shortcut where the model learns to guess quickly; a testable extension would be to randomize target labels and check whether reasoning still tracks the correct cues.
  • Editorial inference: the same GRPO-plus-binary-label recipe is a natural candidate for other visual anomaly tasks such as deepfake detection or document tampering, where attack types are open-ended and textual explanations are expensive.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. This paper proposes a reinforcement fine-tuning framework for cross-domain face anti-spoofing. The method uses Qwen2.5-VL-7B-Instruct as the base model and optimizes it with GRPO using three verifiable rewards: a format reward, a class-consistent reward based on binary real/spoof labels, and a reasoning-consistent reward based on reasoning length. The authors evaluate on four cross-domain protocols (CeFa to HQ-WMCA, CeFa to SiW-Mv2, CASIA-SURF to HQ-WMCA, CASIA-SURF to SiW-Mv2) and report state-of-the-art HTER results, along with ablations, training-data-volume and sampling-quantity studies, and qualitative reasoning examples. The central claim is that reinforcement fine-tuning with only class labels learns transferable decision policies for unseen attack types, without requiring textual reasoning annotations.

Significance. If the empirical claims hold, this is a useful contribution: it demonstrates that binary-label verifiable rewards can adapt a multimodal large language model to face anti-spoofing while producing interpretable reasoning, and it provides a direct comparison against supervised fine-tuning on the same base model in Table 2. The reward design is simple and the ablations are informative. However, the significance is currently limited by unresolved evaluation-protocol concerns, missing uncertainty quantification, and a reward that may encourage verbosity rather than improved reasoning. The central state-of-the-art claim is plausible but not yet rigorously established.

major comments (4)
  1. [Training and Inference Process / Table 1] The central SOTA claim rests on an uncontrolled comparison of decision rules. The paper states in the Training and Inference Process section that 'we do not adopt traditional threshold-based evaluation methods; instead, we directly assess the correctness of the prediction by comparing whether the predicted class is equal to the ground truth class.' Ours is therefore a threshold-free text classification rule, while every baseline in Table 1 (MS-LBP, Color texture, CNN, Flip, FoundPAD) is a score-based classifier whose HTER depends on a decision threshold. The Comparison Methods paragraph does not specify how baseline thresholds were selected: whether on source-domain validation data, on target-domain test data as an oracle/EER, or from original publications. Without this information, the reported improvements of 18.48%, 60.75%, 18.17%, and 28.28% over the second-best method are not verifiable. Please provide the exact threshold-setting protocol for each baseline and, ideally, evaluate Ours under the same threshold-based HTER procedure as well as the threshold-free rule.
  2. [Experiments / Tables 1-3, Figures 4-5] No uncertainty quantification is reported for any result. Tables 1, 2, and 3 and Figures 4 and 5 appear to report a single run per configuration, with no repeated seeds, standard deviations, or statistical tests. Given that reinforcement fine-tuning is stochastic and that several HTER differences are small (e.g., Table 2, CeFa to SiW-Mv2: Ours 8.89% vs SFT 9.76%), the reported differences may fall within run-to-run noise. Please report multiple seeds with means and variances, and apply a significance test or confidence intervals for the headline comparisons.
  3. [Eq. (5)] The reasoning-consistent reward explicitly rewards longer reasoning when the class prediction is correct and penalizes longer reasoning when it is incorrect. As written, this reward can be maximized by verbose correct responses and terse incorrect responses, without any direct measure of whether the reasoning is valid or causally supports the decision. The paper's mechanism story that the model 'distills highly generalizable decision-making rules' is therefore not directly supported by Rres; the experiments only measure classification accuracy, not reasoning quality. Please either measure reasoning quality directly (e.g., human or automated evaluation of the generated rationales) or weaken the claim about distilled decision rules, and consider a reward that penalizes irrelevant or unsupported reasoning rather than length per se.
  4. [Table 1 / Comparison Methods] The comparison in Table 1 mixes model families: Ours uses Qwen2.5-VL-7B-Instruct, a large multimodal model, while the strongest baselines (Flip, FoundPAD) use smaller or specialized models. Table 2 partially controls for this by comparing Qwen-based SFT and RFT, and that comparison is useful evidence. However, the abstract and conclusion state 'state-of-the-art' based on Table 1, where the gain may partly reflect base-model capacity rather than the proposed reinforcement fine-tuning method. Please make this confound explicit, and consider adding a Qwen2.5-VL baseline evaluated under the same threshold-free protocol, or clarify that the SOTA statement is relative to previously published methods with different base models.
minor comments (4)
  1. [Eq. (5)] The notation is inconsistent: the equation uses oi in the text but defines o_i for the policy group. Please unify the notation and explicitly define len(·) and L.
  2. [Ablation Study and Visualization Analysis] The sentence 'When the format reward Rformat and reasoning consistent reward Rres are removed, the HTER metric increases by 16.89% and 24.39%, respectively' is ambiguous: these percentages appear to be computed relative to the ablated HTER values (3.20/18.95 and 5.08/20.83) rather than relative to the full method's HTER of 15.75%. Please state whether the increases are absolute or relative, and to which reference value.
  3. [Figure 6] Figure 6 is very crowded: the attack-type labels, colored bars, and numeric values are difficult to read, especially for SiW-Mv2. Please enlarge fonts, use separate panels per protocol, or provide a table with the per-attack error rates.
  4. [Implementation details] Several GRPO hyperparameters named in Eq. (1) (epsilon, beta, and the reference model) are not reported. Please list all hyperparameters, including the KL penalty coefficient, clipping coefficient, number of GRPO iterations, and the training dataset sizes after filtering.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: reward signals are inputs, target-domain HTER is an independent empirical result, and the cited prior work is not load-bearing.

full rationale

The paper's claimed derivation is an empirical RL fine-tuning pipeline, not a reduction. The class consistent reward (Eq. 4), reasoning consistent reward (Eq. 5), and format reward (Eq. 3) are combined in Eq. 6 and optimized through GRPO (Eqs. 1-2); these rewards are defined from source-domain labels as optimization objectives. The evaluation computes FRR, FAR, and HTER on unseen target domains, which are not entailed by the reward definitions. The target-domain results are therefore an empirical outcome, not an artifact of construction. The only self-citation (Jiang et al. 2024) is used for background on open-set single-domain generalization and is not load-bearing. The paper's threshold-free evaluation statement ('we do not adopt traditional threshold-based evaluation methods; instead, we directly assess the correctness of the prediction by comparing whether the predicted class is equal to the ground truth class') creates a potential comparability concern for the Table 1 baselines, whose HTER values are threshold-dependent, but this is an external-validity and fairness issue for the SOTA claim rather than circularity in the derivation chain. Hence no circular step is identified.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The ledger includes the main hyperparameters the paper sets by hand, the borrowed RL framework, and the domain assumptions about source labels and the base model. No new entities are invented. The largest epistemic cost is the assumption that binary labels plus reward shaping transfer to unseen attack types, which is empirically asserted rather than derived.

free parameters (6)
  • Expected maximum reasoning length L = 1200
    Used in Eq. (5) to reward or penalize reasoning length; chosen by hand and not ablated except through the gating function.
  • Policy group size N = 6
    Number of sampled responses per image in GRPO; Fig. 5 shows performance depends on N.
  • GRPO clipping coefficient epsilon = unspecified
    Appears in Eq. (1); value is not reported in Implementation Details.
  • GRPO KL penalty coefficient beta = unspecified
    Appears in Eq. (1); value is not reported in Implementation Details.
  • Learning rate = 5e-6
    Given in Implementation Details; standard training hyperparameter.
  • Batch size = 6
    Given in Implementation Details; standard training hyperparameter.
assumptions (5)
  • standard math GRPO objective (Eq. 1) is a valid policy optimization algorithm and its standard implementation supports the described rewards.
    Lifted from DeepSeekMath (Shao et al. 2024), not re-derived in this paper.
  • domain assumption Binary labels 'real' and 'fake' in the source domain are correct and sufficient to learn transferable decision rules.
    The rewards rely entirely on ground-truth labels; no independent verification that these labels are transferable.
  • domain assumption Qwen2.5-VL-7B-Instruct provides a strong visual prior, so RL fine-tuning with limited source data yields generalizable behavior.
    The paper selects this base model and does not test smaller or different architectures; performance may be specific to this model.
  • ad hoc to paper The reasoning-consistency reward (Eq. 5) encourages useful reasoning: longer correct reasoning is better, longer wrong reasoning is worse.
    This heuristic is introduced for this task without independent evidence that reasoning length correlates with generalizable decision rules.
  • domain assumption The target domains and source domains are disjoint in attack type; the evaluation protocols construct significant covariate and semantic shift.
    Relies on dataset composition and prior works; no quantitative shift measure is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Exploring Task-Solving Paradigm for Generalized Cross-Domain Face Anti-Spoofing via Reinforcement Fine-Tuning." pith.science (2026). https://pith.science/paper/LVETOGHG

@misc{pith2026250621895,
  author       = {Pith},
  title        = {Pith review of: Exploring Task-Solving Paradigm for Generalized Cross-Domain Face Anti-Spoofing via Reinforcement Fine-Tuning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LVETOGHG}},
  note         = {Machine review of arXiv:2506.21895}
}
read the original abstract

Recently the emergence of novel presentation attacks has drawn increasing attention to face anti-spoofing. However, existing methods tend to memorize data patterns from the training set, resulting in poor generalization to unknown attack types across different scenarios and limited interpretability. To address these challenges, this paper presents a reinforcement fine-tuning-based face anti-spoofing method that stimulates the capabilities of multimodal large language models to think and learn how to solve the anti-spoofing task itself, rather than relying on the memorization of authenticity patterns. We design verifiable class consistent reward and reasoning consistent reward, and employ a GRPO-based optimization strategy to guide the model in exploring reasoning policies from multiple perspectives to maximize expected rewards. As a result, through iterative trial-and-error learning while retaining only high-reward trajectories, the model distills highly generalizable decision-making rules from the extensive solution space to effectively address cross-domain face anti-spoofing tasks. Extensive experimental results demonstrate that our method achieves state-of-the-art cross-domain generalization performance. It generalizes well to diverse unknown attack types in unseen target domains while providing interpretable reasoning for its authenticity decisions without requiring labor-intensive textual annotations for training.

Figures

Figures reproduced from arXiv: 2506.21895 by the authors.

Figure 1
Figure 1. In contrast to supervised fine-tuning, which relies [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the reinforcement fine-tuning framework for generalized cross-scenario face anti-spoofing. The frame [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Sample attacks in the source and target domains. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Performance variation with different training data [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Performance variation with different sampling [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Error rate(%↓) visualization results of various type of faces. The CeFa and CASIA-SURF datasets are the source domains and the HQ-WMCA and SiW-Mv2 datasets are the unseen target domains, respectively. The image appears to be a highly detailed 3D model of a human face, …
Figure 7
Figure 7. Figure 7: Reasoning Result Visualization under the protocol [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 45 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Bai, S.; Chen, K.; Liu, X.; Wang, J.; Ge, W.; Song, S.; Dang, K.; Wang, P.; Wang, S.; Tang, J.; et al. 2025. Qwen2. 5-vl technical report. arXiv preprint arXiv:2502.13923

    arXiv 2025

  4. [4]

    Boulkenafet, Z.; Komulainen, J.; and Hadid, A. 2016. Face spoofing detection using colour texture analysis. IEEE Transactions on Information Forensics and Security, 11(8): 1818--1830

    work page 2016

  5. [5]

    Cai, R.; Li, Z.; Wan, R.; Li, H.; Hu, Y.; and Kot, A. C. 2022. Learning meta pattern for face anti-spoofing. IEEE Transactions on Information Forensics and Security, 17: 1201--1213

    work page 2022

  6. [6]

    Cai, R.; Soh, C.; Yu, Z.; Li, H.; Yang, W.; and Kot, A. C. 2024. Towards Data-Centric Face Anti-spoofing: Improving Cross-Domain Generalization via Physics-Based Data Synthesis. International Journal of Computer Vision, 1--22

    work page 2024

  7. [7]

    Chen, G.; Xie, W.; Lin, D.; Liu, Y.; and Wang, M. 2025. mmFAS: Multimodal Face Anti-Spoofing Using Multi-Level Alignment and Switch-Attention Fusion. In Association for the Advancement of Artificial Intelligence, volume 39, 58--66

    work page 2025

  8. [8]

    V.; Levine, S.; and Ma, Y

    Chu, T.; Zhai, Y.; Yang, J.; Tong, S.; Xie, S.; Schuurmans, D.; Le, Q. V.; Levine, S.; and Ma, Y. 2025. Sft memorizes, rl generalizes: A comparative study of foundation model post-training. arXiv preprint arXiv:2501.17161

    arXiv 2025

Show all 54 references
  1. [9]

    A.; and Nugroho, A

    Dharmawan, D. A.; and Nugroho, A. S. 2024. Towards Deep Face Spoofing: Taxonomy, Recent Advances, and Open Challenges. IEEE Transactions on Biometrics, Behavior, and Identity Science

    2024

  2. [10]

    Dong, X.; Liu, H.; Cai, W.; Lv, P.; and Yu, Z. 2021. Open set face anti-spoofing in unseen attacks. In ACM International Conference on Multimedia, 4082--4090

    2021

  3. [11]

    Fang, H.; Liu, A.; Jiang, N.; Lu, Q.; Zhao, G.; and Wan, J. 2024. VL-FAS: Domain Generalization via Vision-Language Model For Face Anti-Spoofing. In IEEE International Conference on Acoustics, Speech and Signal Processing, 4770--4774

    2024

  4. [12]

    Ge, X.; Liu, X.; Yu, Z.; Shi, J.; Qi, C.; Li, J.; and K \"a lvi \"a inen, H. 2024. Difffas: face anti-spoofing via generative diffusion models. In European Conference on Computer Vision, 144--161

    2024

  5. [13]

    Guo, J.; Liu, A.; Diao, Y.; Zhang, J.; Ma, H.; Zhao, B.; Hong, R.; and Wang, M. 2025. Domain Generalization for Face Anti-spoofing via Content-aware Composite Prompt Engineering. arXiv preprint arXiv:2504.04470

    2025 arXiv

  6. [14]

    Guo, J.; Liu, H.; Luo, Y.; Hu, X.; Zou, H.; Zhang, Y.; Liu, H.; and Zhao, B. 2024. Style-conditional prompt token learning for generalizable face anti-spoofing. In ACM International Conference on Multimedia, 994--1003

    2024

  7. [15]

    Guo, X.; Liu, Y.; Jain, A.; and Liu, X. 2022. Multi-domain learning for updating face anti-spoofing models. In European Conference on Computer Vision, 230--249

    2022

  8. [16]

    Heusch, G.; George, A.; Geissb \"u hler, D.; Mostaani, Z.; and Marcel, S. 2020. Deep models and shortwave infrared information to detect face presentation attacks. IEEE Transactions on Biometrics, Behavior, and Identity Science, 2(4): 399--409

    2020

  9. [17]

    Hu, C.; Zhang, K.-Y.; Yao, T.; Ding, S.; and Ma, L. 2024 a . Rethinking generalizable face anti-spoofing via hierarchical prototype-guided distribution refinement in hyperbolic space. In IEEE Conference on Computer Vision and Pattern Recognition, 1032--1041

    2024

  10. [18]

    Hu, X.; Liu, H.; Yuan, H.; Fu, Z.; Luo, Y.; Zhang, N.; Zou, H.; Gan, J.; and Zhang, Y. 2024 b . Fine-grained prompt learning for face anti-spoofing. In ACM International Conference on Multimedia, 7619--7628

    2024

  11. [19]

    Huang, P.-K.; Chiang, C.-H.; Chen, T.-H.; Chong, J.-X.; Liu, T.-L.; and Hsu, C.-T. 2024. One-Class Face Anti-spoofing via Spoof Cue Map-Guided Feature Learning. In IEEE Conference on Computer Vision and Pattern Recognition, 277--286

    2024

  12. [20]

    Huang, P.-K.; Chong, J.-X.; Chiang, C.-H.; Chen, T.-H.; Liu, T.-L.; and Hsu, C.-T. 2025. SLIP: Spoof-aware one-class face anti-spoofing with language image pretraining. In Association for the Advancement of Artificial Intelligence, volume 39, 3697--3706

    2025

  13. [21]

    Huang, P.-K.; Lu, C.-Y.; Chang, S.-J.; Chong, J.-X.; and Hsu, C.-T. 2023. Test-Time Adaptation for Robust Face Anti-Spoofing. In British Machine Vision Conference, 379--380

    2023

  14. [22]

    Jia, Y.; Zhang, J.; and Shan, S. 2021. Dual-Branch Meta-Learning Network With Distribution Alignment for Face Anti-Spoofing. IEEE Transactions on Information Forensics and Security, 17: 138--151

    2021

  15. [23]

    Jia, Y.; Zhang, J.; Shan, S.; and Chen, X. 2020. Single-side domain generalization for face anti-spoofing. In IEEE Conference on Computer Vision and Pattern Recognition, 8484--8493

    2020

  16. [24]

    Jia, Y.; Zhang, J.; Shan, S.; and Chen, X. 2021. Unified unsupervised and semi-supervised domain adaptation network for cross-scenario face anti-spoofing. Pattern Recognition, 115: 107888

    2021

  17. [25]

    Jiang, F.; Li, Q.; Wang, W.; Ren, M.; Shen, W.; Liu, B.; and Sun, Z. 2024. Open-Set Single-Domain Generalization for Robust Face Anti-Spoofing. International Journal of Computer Vision, 132(11): 5151--5172

    2024

  18. [26]

    Kong, Z.; Zhang, W.; Wang, T.; Zhang, K.; Li, Y.; Tang, X.; and Luo, W. 2024. Dual teacher knowledge distillation with domain alignment for face anti-spoofing. IEEE Transactions on Circuits and Systems for Video Technology

    2024

  19. [27]

    M.; and Woo, S

    Le, B. M.; and Woo, S. S. 2024. Gradient alignment for cross-domain face anti-spoofing. In IEEE Conference on Computer Vision and Pattern Recognition, 188--199

    2024

  20. [28]

    Li, H.; He, P.; Wang, S.; Rocha, A.; Jiang, X.; and Kot, A. C. 2018. Learning Generalized Deep Feature Representation for Face Anti-Spoofing. IEEE Transactions on Information Forensics and Security, 13(10): 2639--2652

    2018

  21. [29]

    K.; and Xing, Y

    Li, Z.; Zhao, T.; Xu, X.; Zhang, Z.; Li, Z.; Chen, X.; Zhang, Q.; Bergamo, A.; Jain, A. K.; and Xing, Y. 2025. Optimal Transport-Guided Source-Free Adaptation for Face Anti-Spoofing. In Computer Vision and Pattern Recognition Conference, 24351--24363

    2025

  22. [30]

    Lin, X.; Liu, A.; Yu, Z.; Cai, R.; Wang, S.; Yu, Y.; Wan, J.; Lei, Z.; Cao, X.; and Kot, A. 2025. Reliable and Balanced Transfer Learning for Generalized Multimodal Face Anti-Spoofing. IEEE Transactions on Pattern Analysis and Machine Intelligence

    2025

  23. [31]

    Liu, A.; Tan, Z.; Wan, J.; Escalera, S.; Guo, G.; and Li, S. Z. 2021. Casia-surf cefa: A benchmark for multi-modal cross-ethnicity face anti-spoofing. In IEEE Winter Conference on Applications of Computer Vision, 1179--1187

    2021

  24. [32]

    Liu, A.; Xue, S.; Gan, J.; Wan, J.; Liang, Y.; Deng, J.; Escalera, S.; and Lei, Z. 2024. CFPL-FAS: Class Free Prompt Learning for Generalizable Face Anti-spoofing. In Conference on Computer Vision and Pattern Recognition, 222--232

    2024

  25. [33]

    Liu, S.-Q.; Wang, Q.; and Yuen, P. C. 2024. Bottom-up domain prompt tuning for generalized face anti-spoofing. In European Conference on Computer Vision, 170--187

    2024

  26. [34]

    Liu, Y.; Chen, Y.; Dai, W.; Gou, M.; Huang, C.-T.; and Xiong, H. 2022. Source-free domain adaptation with contrastive domain alignment and self-supervised exploration for face anti-spoofing. In European Conference on Computer Vision, 511--528

    2022

  27. [35]

    Liu, Y.; Li, Z.; and Wu, L. 2025. Dual Consistency Regularization for Generalized Face Anti-Spoofing. IEEE Transactions on Information Forensics and Security

    2025

  28. [36]

    Ma, Y.; Qian, J.; Li, J.; and Yang, J. 2024. Dual feature disentanglement for face anti-spoofing. Pattern Recognition, 155: 110656

    2024

  29. [37]

    a \"a tt \

    M \"a \"a tt \"a , J.; Hadid, A.; and Pietik \"a inen, M. 2011. Face spoofing detection from single images using micro-texture analysis. In IEEE International Joint Conference on Biometrics, 1--7

    2011

  30. [38]

    Mu, L.; Bai, J.; He, X.; Ye, J.; Liang, X.; Yang, Y.; Zhuang, J.; and Hu, H. 2023. TeG-DG: Textually Guided Domain Generalization for Face Anti-Spoofing. arXiv preprint arXiv:2311.18420

    2023 arXiv

  31. [39]

    Narayan, K.; and Patel, V. M. 2024. Hyp-OC: Hyperbolic One Class Classification for Face Anti-Spoofing. arXiv preprint arXiv:2404.14406

    2024 arXiv

  32. [40]

    Ozgur, G.; Caldeira, E.; Chettaoui, T.; Boutros, F.; Ramachandra, R.; and Damer, N. 2025. FoundPAD: Foundation Models Reloaded for Face Presentation Attack Detection. arXiv preprint arXiv:2501.02892

    2025 arXiv

  33. [41]

    Shao, R.; Lan, X.; Li, J.; and Yuen, P. C. 2019. Multi-adversarial discriminative deep domain generalization for face presentation attack detection. In IEEE Conference on Computer Vision and Pattern Recognition, 10023--10031

    2019

  34. [42]

    Shao, Z.; Wang, P.; Zhu, Q.; Xu, R.; Song, J.; Bi, X.; Zhang, H.; Zhang, M.; Li, Y.; Wu, Y.; et al. 2024. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. arXiv preprint arXiv:2402.03300

    2024 arXiv

  35. [43]

    Srivatsan, K.; Naseer, M.; and Nandakumar, K. 2023. FLIP: Cross-domain Face Anti-spoofing with Language Guidance. In International Conference on Computer Vision, 19685--19696

    2023

  36. [44]

    Wang, H.; Shi, Y.; Tao, Z.; Gao, Y.; Zhang, L.; Lin, X.; Feng, J.; Yuan, X.; Yu, Z.; and Cao, X. 2025. FaceShield: Explainable Face Anti-Spoofing with Multimodal Large Language Models. arXiv preprint arXiv:2505.09415

    2025

  37. [45]

    Wang, K.; Zhang, G.; Yue, H.; Liang, Y.; Huang, M.; Zhang, G.; Han, J.; Ding, E.; and Wang, J. 2024 a . CSDG-FAS: Closed-Space Domain Generalization for Face Anti-spoofing. International Journal of Computer Vision, 132(11): 4866--4879

    2024

  38. [46]

    Wang, X.; Zhang, K.-Y.; Yao, T.; Zhou, Q.; Ding, S.; Dai, P.; and Ji, R. 2024 b . TF-FAS: twofold-element fine-grained semantic guidance for generalizable face anti-spoofing. In European Conference on Computer Vision, 148--168

    2024

  39. [47]

    Wang, Z.; Wang, Z.; Yu, Z.; Deng, W.; Li, J.; Gao, T.; and Wang, Z. 2022. Domain generalization via shuffled style assembly for face anti-spoofing. In IEEE Conference on Computer Vision and Pattern Recognition, 4123--4133

    2022

  40. [48]

    Yang, J.; Lei, Z.; and Li, S. Z. 2014. Learn convolutional neural network for face anti-spoofing. arXiv preprint arXiv:1408.5601

    2014 arXiv

  41. [49]

    Yang, J.; Yu, Z.; Ni, X.; He, J.; and Li, H. 2024. Generalized Face Anti-spoofing via Finer Domain Partition and Disentangling Liveness-irrelevant Factors. In European Conference on Artificial Intelligence, 274--281

    2024

  42. [50]

    Yu, Z.; Qin, Y.; Li, X.; Zhao, C.; Lei, Z.; and Zhao, G. 2023. Deep learning for face anti-spoofing: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(05): 5609--5631

    2023

  43. [51]

    Zhang, D.; Du, Z.; Li, J.; Zhu, L.; and Shen, H. T. 2024 a . Domain-Adaptive Energy-Based Models for Generalizable Face Anti-Spoofing. IEEE Transactions on Multimedia

    2024

  44. [52]

    Zhang, G.; Wang, K.; Yue, H.; Liu, A.; Zhang, G.; Yao, K.; Ding, E.; and Wang, J. 2025. Interpretable Face Anti-Spoofing: Enhancing Generalization with Multimodal Large Language Models. arXiv preprint arXiv:2501.01720

    2025 arXiv

  45. [53]

    Zhang, J.; Huang, J.; Jin, S.; and Lu, S. 2024 b . Vision-language models for vision tasks: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence

    2024

  46. [54]

    Zhang, S.; Wang, X.; Liu, A.; Zhao, C.; Wan, J.; Escalera, S.; Shi, H.; Wang, Z.; and Li, S. Z. 2019. A dataset and benchmark for large-scale multi-modal face anti-spoofing. In IEEE Conference on Computer Vision and Pattern Recognition, 919--928

    2019

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.