Pith. sign in

REVIEW 3 major objections 5 minor 55 references

Frequency Matters: Explaining Biases of Face Recognition in the Frequency Domain

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read The paper claims that ethnicity bias in face recognition is visible in the frequency bands a model relies on, and that this spectral signature grows with training-data bias.

desk verdict A new and sensible question about frequency and ethnicity bias, but the evidence is visual, confounded, and not yet convincing. read the letter →

arxiv 2501.16896 v1 pith:J2CTRTDW submitted 2025-01-28 cs.CV

Marco Huber , Fadi Boutros , Naser Damer This is my paper · ORCID
classification cs.CV
keywords facerecognitionethnicitybiasfrequencydomainexplainabilityimportanceverificationamplification
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

This paper argues that a face recognition model's ethnicity bias has a measurable signature in the frequency domain: the model weighs different frequency bands differently depending on the ethnicity of the faces being compared. Using frequency masking, the authors compute a per-band importance score for African, Asian, Caucasian, and Indian verification pairs and compare these scores across seven models. They find that low frequencies are generally more important for African samples and high frequencies more important for Asian samples, and that these differences grow when a model is trained with one ethnicity removed from its data. This offers a non-semantic, quantitative explanation of ethnicity bias, complementary to earlier explanations based on hairstyle, facial hair, makeup, or illumination.

What carries the argument

The operative mechanism is a black-box frequency-masking explanation. Each face image is transformed by the discrete Fourier transform, the spectrum is divided into disjoint frequency bands of width 4 by Euclidean distance, each band is zeroed out in turn, and the image is transformed back; the absolute change in the model's verification similarity score is that band's importance, normalized so the bands sum to 1. Averaging these importance values per ethnicity and per model, and ranking the bands categorically, lets the authors compare which spectral regions drive recognition for each ethnic group and how the pattern shifts with bias.

What would settle it

Train the same face recognition models on data augmented with frequency-masked images, then rerun the band-masking analysis: if the ethnicity-dependent frequency importance differences disappear, the original observation was an out-of-distribution artifact rather than a genuine bias signature.

Watch

Extended reading notes

Core claim

The central claim is that different frequency bands carry different importance for face recognition decisions depending on the ethnicity of the input face pair, and that this ethnicity-dependent frequency importance is amplified in more biased models. Across the baseline model and the intentionally biased models, African faces show higher reliance on low-frequency bands than other ethnicities, while Asian faces show higher reliance on high-frequency bands; the two pre-trained models also show ethnicity-dependent differences, though less pronounced. Removing one ethnicity from training systematically shifts the importance profile for that group, most visibly in low-frequency bands. The paper concludes that frequency-based explanations expose a source of ethnicity bias that is invisible to spatial-domain semantic explanations.

Load-bearing premise

The conclusion rests on the assumption that removing a frequency band and measuring the drop in similarity reveals how much the model relies on that band, rather than simply how surprised the model is by masked images.

Editorial extensions

If this is right

  • For a given model, the relative importance ranking of frequency bands differs by ethnicity, with a recurring pattern of low-frequency dominance for African faces and high-frequency prominence for Asian faces.
  • Models trained without one ethnicity show larger between-ethnicity gaps in frequency importance, mainly at low-frequency bands, compared with the baseline model.
  • Excluding an ethnicity from training sharply lowers verification accuracy on that group, and the frequency-importance measurement tracks that drop: the affected group's low-frequency importance increases in most cases.
  • Two pre-trained models trained on an unbalanced public dataset also display ethnicity-dependent frequency importance, but with smaller between-ethnicity contrasts than the deliberately biased models.
  • Because the ranking method is categorical, the finding describes trends in spectral reliance rather than exact magnitudes; nominal importance values are reported in supplementary material.

Reading between the lines

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

  • If the frequency-importance signature is a genuine correlate of bias, it could be turned into an audit metric for face recognition models, detecting disparities even when overall accuracy looks balanced.
  • Because ethnicity labels are coarse, a natural check is to split the same data by measured skin tone or image brightness; if the low-frequency effect tracks photometric statistics rather than demographic categories, the mechanism is more about image properties than identity groups.
  • A debiasing experiment could test whether training with frequency-band dropout or a regularization that equalizes per-ethnicity frequency importance reduces verification disparities, turning the diagnostic into a mitigation.
  • The pairwise analysis could be sharpened by comparing the same face pair across models instead of comparing aggregated distributions, which would remove pair-level variance and isolate the bias-driven shift.
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

3 major / 5 minor

Summary. The paper investigates ethnicity bias in face recognition (FR) by analyzing how much individual frequency bands contribute to verification decisions. Using the frequency-masking explanation method of Ref. [26], the authors compute a per-band importance score h_b = |sc(N(j),N(k)) - sc(N(j_{M,b}),N(k_{M,b}))| (Eq. 3) for two pre-trained FR models (AdaFace, ElasticFace-Cos) and five ResNet-34 models trained on BUPT-Balanceface (one baseline and four models each trained with one ethnicity excluded). Evaluations on the RFW dataset are used to compare frequency-importance rankings and mean importance distributions across Asian, African, Caucasian, and Indian pairs. The paper's central claim is that different frequencies are important to FR models depending on the ethnicity of the samples, and that this effect is amplified in more biased models, particularly in low-frequency bands.

Significance. If substantiated, the result would be a valuable non-semantic, measurable correlate of demographic bias in face recognition, complementing semantic explanations such as hairstyle or facial hair. The paper has several strengths: it uses standard bias evaluation datasets and protocols (RFW, BFW in the supplementary), considers multiple models with different architectures and losses, follows a published frequency-explanation method, and explicitly acknowledges the out-of-distribution limitation of perturbation-based explanations in Section 3.1. However, the current evidence is descriptive: the central claims rest on visual inspection of plots without significance tests, and the perturbation-magnitude confound identified below is not controlled. The significance is therefore conditional on additional analysis rather than established by the present manuscript.

major comments (3)
  1. [§5.2–§5.3, Figs. 1–5] The central claim that frequency importance differs by ethnicity and increases with model bias is currently supported only by visual inspection of mean ranking plots (Fig. 1) and mean importance bar charts (Figs. 2–5). The text itself acknowledges the large variation between pairs and defers standard deviation plots to the supplementary material (§5.3). Please add inferential statistics over the 6,000-pair ethnicity protocols (e.g., bootstrap confidence intervals for P_{b,e}, permutation tests for ethnicity differences, and paired tests for baseline-vs-biased model differences) and report effect sizes. Without such tests, the observed between-ethnicity differences and their increase under bias cannot be distinguished from sampling variation.
  2. [§3.1, Eq. (3), §4.4] The frequency-importance measure in Eq. (3) is not calibrated across ethnicities. Because the DFT bands are defined by Euclidean distance in the frequency grid (§4.4), masking frequency band b always removes the same coordinate set but not the same spectral energy for every image; if ethnic groups differ in low-frequency content (due to illumination, skin-tone statistics, or acquisition differences), Eq. (3) will produce group-dependent importance values even for a model that is entirely ethnicity-agnostic. The out-of-distribution caveat in §3.1 does not address this magnitude confound. Please quantify the per-band spectral energy removed per ethnicity and add a control condition (e.g., equal-energy masking, random-phase perturbation, or pseudo-images with matched spectra) to show that the results reflect model reliance rather than perturbation magnitude.
  3. [§5.3, Table 1, Eq. (3)] The comparison supporting the 'more biased models' conclusion is confounded by overall verification accuracy differences. Table 1 shows that each biased model drops sharply on the excluded ethnicity (e.g., M_Afr: 80.25% vs. 92.92% on African), so the absolute similarity differences in Eq. (3) are not comparable across models because the baseline similarity scores differ substantially. Please report relative importance (e.g., the drop divided by the unperturbed similarity score) or otherwise control for the baseline score distribution, and perform the baseline-vs-biased comparison on the same image pairs.
minor comments (5)
  1. [§3.1, Eqs. (4)–(5)] Equations (4) and (5) reuse the symbol h_b for the raw, min-scaled, and sum-normalized importance values. Moreover, dividing by min(h_b) before sum-normalization is redundant and can be undefined if the minimum is zero; please introduce distinct symbols such as h_b^{(raw)}, h_b^{(min)}, and h_b^{(norm)}.
  2. [§3.3] The phrase 'rank the obtained mean importance values P_{b,e} for each frequency band b' is ambiguous: it should state explicitly whether the ranking is over ethnicities per band or over bands per ethnicity, and how ties are handled.
  3. [§5.2, Fig. 1] The AdaFace ranking is reported as an exception (high frequencies more important for African faces), but this discrepancy is not discussed. A brief interpretation would help the reader understand whether the central claim applies uniformly across models or only to a subset of them.
  4. [Throughout] There are several typos and grammar issues, including 'its important' (§2.1), 'tent to utilize' (§5.3), 'ethnicitiy' (Eq. 6), and 'bar indicate' (Fig. 3 caption); these should be corrected.
  5. [Supplementary material, BFW] The BFW results are only mentioned as being in the supplementary material; the main text should include a brief summary of whether the BFW findings reproduce the RFW findings.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: the paper reports an empirical measurement of frequency-band importance per ethnicity, and no fitted parameter or definition forces the observed cross-ethnicity differences.

full rationale

The central claim is an empirical observation, not a derivation from a fitted parameter or a self-referential definition. Equation (3) defines band importance as the absolute similarity drop after masking a frequency band; this measure is taken from the authors' earlier work [26], but it is not fitted to the ethnicity-split data, and the per-pair normalization in Eqs. (4) and (5) is a positive monotone rescaling applied identically to all ethnicities, so it cannot by construction produce the reported differences between ethnicity groups. The biased models are constructed by removing one ethnicity subset from training data (Sec. 3.2), which does create performance disparities by design, but the frequency-importance values are separately measured outputs and are not algebraic consequences of the training-set construction. The paper explicitly discloses the out-of-distribution limitation of frequency masking (Sec. 3.1, 'a limitation of choosing this black-box masking approach... is, that out-of-distribution behavior can appear, since the model is not trained on (frequency-) masked data'); that is a validity caveat about whether the importance scores faithfully reflect decision reliance, not a circular step. The absence of statistical tests in the comparative figures is a correctness/robustness concern, but it does not make the claim equivalent to its inputs. The self-citations [26] and [28] supply the measurement tool and the bias-amplification recipe, but those prior works do not assume the paper's target conclusion, so the derivation chain does not reduce to itself.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The central claim rests on a perturbation-based importance measure from the authors' prior work plus several stated assumptions: unbiased models show no frequency-importance pattern, leave-one-out training represents natural bias, and coarse ethnicity labels are meaningful. No new entities are introduced. The only hand-chosen analysis parameter is the frequency band size s=4.

free parameters (1)
  • frequency band size s = 4
    Chosen by hand following Ref. [26] to balance granularity and robustness. The frequency-importance plots and all downstream ethnicity comparisons depend on this resolution.
assumptions (5)
  • standard math DFT and inverse DFT are lossless and preserve all image information.
    Invoked in Eqs. (1) and (2); standard signal processing background.
  • domain assumption The decrease in similarity after removing a frequency band measures the importance of that band to the model's decision.
    Core to Eq. (3), inherited from Ref. [26]. Masking can produce out-of-distribution inputs, so the assumption is not independently validated here.
  • domain assumption An unbiased FR model would show no clear ethnicity-dependent pattern in frequency importance.
    Stated in Section 3.3 as the basis for interpreting any observed pattern as bias-related.
  • domain assumption Training without an ethnicity subset amplifies model bias in a way representative of natural bias.
    Stated in Section 3.2, following Ref. [28]. The amplified-bias models are used to infer that bias increases frequency-importance differences.
  • domain assumption The four coarse ethnicity labels in RFW and BUPT are meaningful categories for measuring bias.
    Used throughout the evaluation. The paper itself acknowledges in Section 2.1 that this oversimplifies a complex, non-binary attribute.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Frequency Matters: Explaining Biases of Face Recognition in the Frequency Domain." pith.science (2026). https://pith.science/paper/J2CTRTDW

@misc{pith2026250116896,
  author       = {Pith},
  title        = {Pith review of: Frequency Matters: Explaining Biases of Face Recognition in the Frequency Domain},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J2CTRTDW}},
  note         = {Machine review of arXiv:2501.16896}
}
read the original abstract

Face recognition (FR) models are vulnerable to performance variations across demographic groups. The causes for these performance differences are unclear due to the highly complex deep learning-based structure of face recognition models. Several works aimed at exploring possible roots of gender and ethnicity bias, identifying semantic reasons such as hairstyle, make-up, or facial hair as possible sources. Motivated by recent discoveries of the importance of frequency patterns in convolutional neural networks, we explain bias in face recognition using state-of-the-art frequency-based explanations. Our extensive results show that different frequencies are important to FR models depending on the ethnicity of the samples.

Figures

Figures reproduced from arXiv: 2501.16896 by the authors.

Figure 1
Figure 1. Relative frequency ranking of the ethnicity for all models. Different pattern can be observed, such as low importance of high frequencies for African (red) or high importance of high frequencies for Asian (orange) in the most cases compared to the other ethnicities [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗
Figure 2
Figure 2. Mean frequency importance distribution based on ethnicity for model M 6 Conclusion In this work, we provided a new perspective on possible causes of ethnicity bias in deep learning-based FR models. Unlike previous works that focus on semantic and visual perspectives in explaining FR bias, we inspect bias in the frequency domain. This was motivated by recent studies pointing out that imperceptible frequency patterns … view at source ↗
Figure 3
Figure 3. Mean frequency importance distribution based on Ethnicity for model MAfr and MAsi. The striped bar indicate the distribution of the baseline M, brightness changes indicate differences between M and the biased models [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Mean frequency importance distribution based on Ethnicity for model MCau and MInd. The striped bar indicate the distribution of the baseline M, brightness changes indicate differences between M and the biased models [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 5
Figure 5. Figure 5: Mean frequency importance distribution based on ethnicity for the two state￾of-the-art models [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 27 canonical work pages

  1. [26]

    Huber, M., Damer, N.: Beyond spatial explanations: Explainable face recognition in the frequency domain (2024),https://arxiv.org/abs/2407.11941 2, 3, 4, 5, 6, 7, 8

    work page Pith review arXiv 2024

  2. [1]

    In: IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPR Workshops 2021, virtual, June 19-25,

    Abello, A.A., Hirata, R., Wang, Z.: Dissecting the high-frequency bias in con- volutional neural networks. In: IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPR Workshops 2021, virtual, June 19-25,

    work page 2021

  3. [2]

    In: Vera-Rodríguez, R., Fiérrez, J., Morales, A

    Acien, A., Morales, A., Vera-Rodríguez, R., Bartolome, I., Fiérrez, J.: Mea- suring the gender and ethnicity bias in deep models for face recognition. In: Vera-Rodríguez, R., Fiérrez, J., Morales, A. (eds.) Progress in Pattern Recog- nition, Image Analysis, Computer Vision, and Applications - 23rd Iberoamerican Congress, CIARP 2018, Madrid, Spain, Novembe...

    work page doi:10.1007/978-3-030-13 2018

  4. [3]

    IEEE Transactions on Informa- tion Forensics and Security pp

    Ahn, K., Lee, S., Han, S., Low, C.Y., Cha, M.: Uncertainty-aware face embedding with contrastive learning for open-set evaluation. IEEE Transactions on Informa- tion Forensics and Security pp. 1–1 (2024). https://doi.org/10.1109/TIFS. 2024.3426973 4

    arXiv 2024

  5. [4]

    In: 31st British Machine Vision Conference 2020, BMVC 2020, Virtual Event, UK, September 7-10, 2020

    Albiero, V., Bowyer, K.W.: Is face recognition sexist? no, gendered hairstyles and biology are. In: 31st British Machine Vision Conference 2020, BMVC 2020, Virtual Event, UK, September 7-10, 2020. BMVA Press (2020),https://www.bmvc2020- conference.com/assets/papers/0905.pdf 2

    work page 2020

  6. [5]

    In: IEEE Winter Appli- cations of Computer Vision Workshops, WACV Workshops 2020, Snowmass Vil- lage, CO, USA, March 1-5, 2020

    Albiero, V., S, K.K., Vangara, K., Zhang, K., King, M.C., Bowyer, K.W.: Anal- ysis of gender inequality in face recognition accuracy. In: IEEE Winter Appli- cations of Computer Vision Workshops, WACV Workshops 2020, Snowmass Vil- lage, CO, USA, March 1-5, 2020. pp. 81–89. IEEE (2020).https://doi.org/10. 1109/WACVW50321.2020.9096947, https://doi.org/10.110...

    arXiv 2020

  7. [12]

    132, 104645 (2023)

    Cascone, L., Pero, C., Proença, H.: Visual and textual explainability for a biometric verificationsystembasedonpiecewisefacialattributeanalysis.ImageVis.Comput. 132, 104645 (2023). https://doi.org/10.1016/J.IMAVIS.2023.104645 , https: //doi.org/10.1016/j.imavis.2023.104645 4

    arXiv 2023

  8. [13]

    Cavazos, J.G., Phillips, P.J., Castillo, C.D., O’Toole, A.J.: Accuracy comparison across face recognition algorithms: Where are we on measuring race bias? IEEE Trans. Biom. Behav. Identity Sci. 3(1), 101–111 (2021). https://doi.org/10. 1109/TBIOM.2020.3027269, https://doi.org/10.1109/TBIOM.2020.3027269 3, 5

    work page arXiv 2021

Show all 55 references
  1. [15]

    IEEE Access12, 34390–34401 (2024)

    DeAndres-Tame, I., Tolosana, R., Vera-Rodríguez, R., Morales, A., Fiérrez, J., Ortega-Garcia, J.: How good is chatgpt at face biometrics? A first look into recog- nition, soft biometrics, and explainability. IEEE Access12, 34390–34401 (2024). https://doi.org/10.1109/ACCESS.202...

    2024

  2. [16]

    In: IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019

    Deng, J., Guo, J., Xue, N., Zafeiriou, S.: Arcface: Additive angular margin loss for deep face recognition. In: IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019. pp. 4690–4699. Computer Vision Foundation / IEEE (2019)...

    2019

  3. [17]

    In: 28th European Signal Processing Conference, EUSIPCO 2020, Amsterdam, Netherlands, January 18- 21, 2021

    Fang, M., Damer, N., Kirchbuchner, F., Kuijper, A.: Demographic bias in pre- sentation attack detection of iris recognition systems. In: 28th European Signal Processing Conference, EUSIPCO 2020, Amsterdam, Netherlands, January 18- 21, 2021. pp. 835–839. IEEE (2020).https://doi...

    2020

  4. [18]

    Pattern Recognit.147, 110002 (2024)

    Fang, M., Yang, W., Kuijper, A., Struc, V., Damer, N.: Fairness in face presenta- tion attack detection. Pattern Recognit.147, 110002 (2024). https://doi.org/ 10.1016/J.PATCOG.2023.110002 , https://doi.org/10.1016/j.patcog.2023. 110002 3 Frequency Matters 17

    2024

  5. [19]

    In: IEEE International Joint Conference on Biometrics, IJCB 2022, Abu Dhabi, United Arab Emirates, October 10-13, 2022

    Fu, B., Damer, N.: Towards explaining demographic bias through the eyes of face recognition models. In: IEEE International Joint Conference on Biometrics, IJCB 2022, Abu Dhabi, United Arab Emirates, October 10-13, 2022. pp. 1–10. IEEE (2022). https://doi.org/10.1109/IJCB54206....

    2022

  6. [20]

    In: 2018 International Conference of the Biometrics Special Interest Group, BIOSIG 2018, Darmstadt, Germany, September 26-28, 2018

    Galbally, J., Haraksim, R., Beslay, L.: Fingerprint quality: a lifetime story. In: 2018 International Conference of the Biometrics Special Interest Group, BIOSIG 2018, Darmstadt, Germany, September 26-28, 2018. LNI, vol. P-282, pp. 1–5. GI / IEEE (2018). https://doi.org/10.239...

    2018

  7. [21]

    In: Leibe, B., Matas, J., Sebe, N., Welling, M

    Guo,Y.,Zhang,L.,Hu,Y.,He,X.,Gao,J.:Ms-celeb-1m:Adatasetandbenchmark for large-scale face recognition. In: Leibe, B., Matas, J., Sebe, N., Welling, M. (eds.) Computer Vision - ECCV 2016 - 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Pa...

    2016 doi

  8. [22]

    https://doi.org/https://doi.org/10.1016/j.imavis.2024.104917 , https://www.sciencedirect.com/science/article/pii/S0262885624000209 3

    Hambarde,K.,Proença,H.:Image-basedhumanre-identification:Whichcovariates are actually (the most) important? Image and Vision Computing 143, 104917 (2024). https://doi.org/https://doi.org/10.1016/j.imavis.2024.104917 , https://www.sciencedirect.com/science/article/pii/S02628856...

    2024

  9. [23]

    In: 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016

    He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016. pp. 770–778. IEEE Computer Society (2016). https://doi.org/10.1109/CVPR.2016...

    2016 doi

  10. [24]

    In: 10th European Workshop on Visual Information Processing, EUVIP 2022, Lisbon, Portugal, September 11-14, 2022

    Herranz, N.M., Galdi, C., Dugelay, J.: Impact of digital face beautification in bio- metrics. In: 10th European Workshop on Visual Information Processing, EUVIP 2022, Lisbon, Portugal, September 11-14, 2022. pp. 1–6. IEEE (2022). https: //doi.org/10.1109/EUVIP53989.2022.992280...

    2022

  11. [27]

    In: Proceedings of the IEEE/CVF Winter Confer- ence on Applications of Computer Vision (WACV)

    Huber, M., Luu, A.T., Boutros, F., Kuijper, A., Damer, N.: Bias and diversity in synthetic-based face recognition. In: Proceedings of the IEEE/CVF Winter Confer- ence on Applications of Computer Vision (WACV). pp. 6215–6226 (January 2024) 3

    2024

  12. [30]

    In: 33rd British Machine Vision Conference 2022, BMVC 2022, London, UK, November 21-24, 2022

    Huber, M., Terhörst, P., Kirchbuchner, F., Damer, N., Kuijper, A.: Stating com- parison score uncertainty and verification decision confidence towards transpar- ent face recognition. In: 33rd British Machine Vision Conference 2022, BMVC 2022, London, UK, November 21-24, 2022. ...

    2022

  13. [31]

    Standard, International Organization for Standard- ization (2024) 2

    ISO/IEC DIS 19795-10: Information technology -Biometric performance testing and reporting - Part 10: Quantifying biometric system performance variation across demographic groups. Standard, International Organization for Standard- ization (2024) 2

    2024

  14. [32]

    IEEE Trans

    Jain, A.K., Deb, D., Engelsma, J.J.: Biometrics: Trust, but verify. IEEE Trans. Biom. Behav. Identity Sci. 4(3), 303–323 (2022). https://doi.org/10.1109/ TBIOM.2021.3115465, https://doi.org/10.1109/TBIOM.2021.3115465 4

    2022

  15. [33]

    In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022, New Orleans, LA, USA, June 18-24, 2022

    Kim,M.,Jain,A.K.,Liu,X.:Adaface:Qualityadaptivemarginforfacerecognition. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022, New Orleans, LA, USA, June 18-24, 2022. pp. 18729–18738. IEEE (2022). https://doi.org/10.1109/CVPR52688.2022.01819, https://d...

    2022

  16. [34]

    IEEE Trans

    Klare, B., Burge, M.J., Klontz, J.C., Bruegge, R.W.V., Jain, A.K.: Face recognition performance: Role of demographic information. IEEE Trans. Inf. Forensics Secur. 7(6), 1789–1801 (2012). https://doi.org/10.1109/TIFS.2012.2214212 , https: //doi.org/10.1109/TIFS.2012.2214212 3, 5

    2012

  17. [35]

    In: IEEE/CVF Winter Conference on Applications of Computer Vision Workshops, WACV 2023 - Workshops, Waikoloa, HI, USA, January 3-7, 2023

    Knoche, M., Teepe, T., Hörmann, S., Rigoll, G.: Explainable model-agnostic sim- ilarity and confidence in face verification. In: IEEE/CVF Winter Conference on Applications of Computer Vision Workshops, WACV 2023 - Workshops, Waikoloa, HI, USA, January 3-7, 2023. pp. 1–8. IEEE ...

    2023

  18. [36]

    ACM Trans

    Lin, Y., Liu, Z.Y., Chen, Y., Wang, Y., Chang, Y., Hsu, W.H.: xcos: An explainable cosine metric for face verification task. ACM Trans. Multim. Comput. Commun. Appl. 17(3s), 112:1–112:16 (2021). https://doi.org/10.1145/3469288 , https: //doi.org/10.1145/3469288 4

    2021 doi

  19. [37]

    In: 2024 IEEE 18th International Conference on Automatic Face and Gesture Recognition (FG)

    Lu, Y., Xu, Z., Ebrahimi, T.: Explainable face verification via feature-guided gra- dient backpropagation. In: 2024 IEEE 18th International Conference on Automatic Face and Gesture Recognition (FG). pp. 1–5 (2024).https://doi.org/10.1109/ FG59268.2024.10581925 2, 4

    2024

  20. [38]

    In: IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2024, Waikoloa, HI, USA, January 3-8, 2024

    Lu, Y., Xu, Z., Ebrahimi, T.: Towards visual saliency explanations of face veri- fication. In: IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2024, Waikoloa, HI, USA, January 3-8, 2024. pp. 4714–4723. IEEE (2024). https://doi.org/10.1109/WACV57701.2024.004...

    2024

  21. [39]

    In: IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition Workshops, CVPR Work- shops 2022, New Orleans, LA, USA, June 19-20, 2022

    Mery, D.: True black-box explanation in facial analysis. In: IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition Workshops, CVPR Work- shops 2022, New Orleans, LA, USA, June 19-20, 2022. pp. 1595–1604. IEEE (2022). https://doi.org/10.1109/CVPRW56347.2022.00166 , h...

    2022

  22. [40]

    In: IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2022, Waikoloa, Frequency Matters 19 HI, USA, January 3-8, 2022

    Mery, D., Morris, B.: On black-box explanation for face verification. In: IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2022, Waikoloa, Frequency Matters 19 HI, USA, January 3-8, 2022. pp. 1194–1203. IEEE (2022). https://doi.org/ 10.1109/WACV51458.2022.00...

    2022

  23. [41]

    Mittal, S., Thakral, K., Majumdar, P., Vatsa, M., Singh, R.: Are face detection models biased? In: 17th IEEE International Conference on Automatic Face and GestureRecognition,FG2023,WaikoloaBeach,HI,USA,January5-8,2023.pp.1–

    2023

  24. [42]

    In: Brömme, A., Busch, C., Dantcheva, A., Raja, K.B., Rathgeb, C., Uhl, A

    Molina, D., Causa, L., Tapia, J.E.: Toward to reduction of bias for gender and eth- nicity from face images using automated skin tone classification. In: Brömme, A., Busch, C., Dantcheva, A., Raja, K.B., Rathgeb, C., Uhl, A. (eds.) BIOSIG 2020 - Proceedings of the 19th Interna...

    2020

  25. [43]

    https://doi.org/10.1109/FG57933.2023.10042564 , https:// doi.org/10.1109/FG57933.2023.10042564 3

    IEEE (2023). https://doi.org/10.1109/FG57933.2023.10042564 , https:// doi.org/10.1109/FG57933.2023.10042564 3

    2023

  26. [44]

    CoRRabs/2208.09500 (2022)

    Neto, P.C., Gonçalves, T., Pinto, J.R., Silva, W., Sequeira, A.F., Ross, A., Cardoso, J.S.: Explainable biometrics in the age of deep learning. CoRRabs/2208.09500 (2022). https://doi.org/10.48550/ARXIV.2208.09500 , https://doi.org/10. 48550/arXiv.2208.09500 1, 3, 4

    2022 doi arXiv

  27. [45]

    In: Damer, N., Gomez-Barrero, M., Raja, K.B., Rathgeb, C., Sequeira, A.F., Todisco, M., Uhl, A

    Neto, P.C., Caldeira, E., Cardoso, J.S., Sequeira, A.F.: Compressed models de- compress race biases: What quantized models forget for fair face recognition. In: Damer, N., Gomez-Barrero, M., Raja, K.B., Rathgeb, C., Sequeira, A.F., Todisco, M., Uhl, A. (eds.) International Con...

    2023

  28. [46]

    Robinson, J.P., Livitz, G., Henon, Y., Qin, C., Fu, Y., Timoner, S.: Face recogni- tion: Too bias, or not too bias? In: 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR Workshops 2020, Seattle, WA, USA, June 14-19, 2020. pp. 1–10. Computer Vision Found...

    2020

  29. [47]

    In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023 - Workshops, Vancouver, BC, Canada, June 17- 24, 2023

    Neto, P.C., Sequeira, A.F., Cardoso, J.S., Terhörst, P.: Pic-score: Probabilistic in- terpretable comparison score for optimal matching confidence in single- and multi- biometric face recognition. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023 - ...

    2023

  30. [48]

    In: IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, 20 M

    Selvaraju, R.R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., Batra, D.: Grad- cam: Visual explanations from deep networks via gradient-based localization. In: IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, 20 M. Huber et al. October 22-29, 201...

    2017 doi

  31. [49]

    In: IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPR Workshops 2019, Long Beach, CA, USA, June 16-20, 2019

    S, K.K., Vangara, K., King, M.C., Albiero, V., Bowyer, K.W.: Characterizing the variability in face recognition accuracy relative to race. In: IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPR Workshops 2019, Long Beach, CA, USA, June 16-20, 2019. pp. ...

    2019

  32. [50]

    IEEE Access12, 1102–1116 (2024).https://doi

    Stragapede, G., Vera-Rodríguez, R., Tolosana, R., Morales, A., Damer, N., Fiér- rez, J., Ortega-Garcia, J.: Keystroke verification challenge (KVC): biometric and fairness benchmark evaluation. IEEE Access12, 1102–1116 (2024).https://doi. org/10.1109/ACCESS.2023.3345452 , https...

    2024

  33. [51]

    In: 2019 IEEE/CVF Interna- tional Conference on Computer Vision, ICCV 2019, Seoul, Korea (South), October 27 - November 2, 2019

    Shi, Y., Jain, A.K.: Probabilistic face embeddings. In: 2019 IEEE/CVF Interna- tional Conference on Computer Vision, ICCV 2019, Seoul, Korea (South), October 27 - November 2, 2019. pp. 6901–6910. IEEE (2019).https://doi.org/10.1109/ ICCV.2019.00700, https://doi.org/10.1109/ICC...

    2019

  34. [52]

    IEEE Transactions on Technology and Society3(1), 16–30 (2022)

    Terhörst, P., Kolf, J.N., Huber, M., Kirchbuchner, F., Damer, N., Moreno, A.M., Fierrez, J., Kuijper, A.: A comprehensive study on face recognition biases beyond demographics. IEEE Transactions on Technology and Society3(1), 16–30 (2022). https://doi.org/10.1109/TTS.2021.3111823 1, 3

    2022

  35. [53]

    In: 2020 IEEE International Joint Conference on Biomet- rics, IJCB 2020, Houston, TX, USA, September 28 - October 1, 2020

    Terhörst, P., Kolf, J.N., Damer, N., Kirchbuchner, F., Kuijper, A.: Face qual- ity estimation and its correlation to demographic and non-demographic bias in face recognition. In: 2020 IEEE International Joint Conference on Biomet- rics, IJCB 2020, Houston, TX, USA, September 2...

    2020

  36. [54]

    https://doi.org/10.1109/IJCB48548.2020.9304865 , https: //doi.org/10.1109/IJCB48548.2020.9304865 3

    IEEE (2020). https://doi.org/10.1109/IJCB48548.2020.9304865 , https: //doi.org/10.1109/IJCB48548.2020.9304865 3

    2020

  37. [55]

    In: 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2020, Seattle, WA, USA, June 13-19, 2020

    Wang, M., Deng, W.: Mitigating bias in face recognition using skewness-aware re- inforcement learning. In: 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2020, Seattle, WA, USA, June 13-19, 2020. pp. 9319–

    2020

  38. [56]

    In: 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR Workshops 2020, Seattle, WA, USA, June 14-19, 2020

    Wang, H., Wang, Z., Du, M., Yang, F., Zhang, Z., Ding, S., Mardziel, P., Hu, X.: Score-cam: Score-weighted visual explanations for convolutional neural networks. In: 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR Workshops 2020, Seattle, WA, USA, Jun...

    2020

  39. [58]

    IEEE Trans

    Wang, M., Zhang, Y., Deng, W.: Meta balanced network for fair face recogni- tion. IEEE Trans. Pattern Anal. Mach. Intell.44(11), 8433–8448 (2022). https: //doi.org/10.1109/TPAMI.2021.3103191 , https://doi.org/10.1109/TPAMI. 2021.3103191 3, 6, 7, 8

    2022

  40. [59]

    In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023 - Workshops, Vancouver, BC, Canada, June 17-24, 2023

    Wu, H., Albiero, V., Krishnapriya, K.S., King, M.C., Bowyer, K.W.: Face recog- nition accuracy across demographics: Shining a light into the problem. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023 - Workshops, Vancouver, BC, Canada, June 17-24, 2...

    2023

  41. [60]

    Neurocomputing429, 215– 244 (2021)

    Wang, M., Deng, W.: Deep face recognition: A survey. Neurocomputing429, 215– 244 (2021). https://doi.org/10.1016/J.NEUCOM.2020.10.081 , https://doi. org/10.1016/j.neucom.2020.10.081 1, 3

    2021 doi

  42. [61]

    In: 2019 IEEE/CVF International Conference on Computer Vision, ICCV 2019, Seoul, Korea (South), Frequency Matters 21 October 27 - November 2, 2019

    Wang, M., Deng, W., Hu, J., Tao, X., Huang, Y.: Racial faces in the wild: Reducing racial bias by information maximization adaptation network. In: 2019 IEEE/CVF International Conference on Computer Vision, ICCV 2019, Seoul, Korea (South), Frequency Matters 21 October 27 - Nove...

    2019

  43. [62]

    CoRRabs/1604.02878 (2016), http://arxiv.org/abs/1604.02878 7

    Zhang, K., Zhang, Z., Li, Z., Qiao, Y.: Joint face detection and alignment us- ing multi-task cascaded convolutional networks. CoRRabs/1604.02878 (2016), http://arxiv.org/abs/1604.02878 7

    2016 arXiv

  44. [64]

    In: IEEE/CVF Winter Conference on Applications of Computer Vision Work- shops, WACVW 2024 - Workshops, Waikoloa, HI, USA, January 1-6, 2024

    Wu, H., Tian, S., Bhatta, A., Öztürk, K., Ricanek, K., Bowyer, K.W.: Fa- cial hair area in face recognition across demographics: Small size, big effect. In: IEEE/CVF Winter Conference on Applications of Computer Vision Work- shops, WACVW 2024 - Workshops, Waikoloa, HI, USA, Ja...

    2024

  45. [65]

    CoRR abs/1411.7923 (2014), http://arxiv.org/abs/1411.7923 7, 8

    Yi, D., Lei, Z., Liao, S., Li, S.Z.: Learning face representation from scratch. CoRR abs/1411.7923 (2014), http://arxiv.org/abs/1411.7923 7, 8

    2014 arXiv

  46. [2021]

    pp. 863–871. Computer Vision Foundation / IEEE (2021). https : / / doi.org/10.1109/CVPRW53098.2021.00096 , https://openaccess.thecvf.com/ content/CVPR2021W/UG2/html/Abello_Dissecting_the_High-Frequency_Bias_ in_Convolutional_Neural_Networks_CVPRW_2021_paper.html 2, 3, 4

    2021

  47. [9328]

    Computer Vision Foundation / IEEE (2020).https://doi.org/10.1109/ CVPR42600.2020.00934, https://openaccess.thecvf.com/content_CVPR_2020/ html/Wang_Mitigating_Bias_in_Face_Recognition_Using_Skewness- Aware_ Reinforcement_Learning_CVPR_2020_paper.html 3, 7

    2020

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.