Pith. sign in

REVIEW 4 major objections 7 minor 45 references

MoDE: Mixture of Diffusion Experts for Any Occluded Face Recognition

T0 review · 4 major / 7 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read This paper claims that adaptively weighting multiple diffusion-repainted versions of an occluded face, fused in the decision space by a learned identity gate, improves recognition over using the occluded image alone.

desk verdict A reasonable plug-and-play diffusion-repaint ensemble for occluded face recognition, but the paper's central 'outperforms all' claim is false on its own tables and the mask-availability assumption is left unspecified. read the letter →

arxiv 2505.04306 v1 pith:QAW6GMUJ submitted 2025-05-07 cs.CV

Qiannan Fan , Zhuoyang Li , Jitong Li , Chenyang Cao This is my paper · ORCID
classification cs.CV
keywords occludedfacerecognitionMixtureofDiffusionExpertsrepaintingidentitygatingdecision-spacefusionmaskocclusionplug-and-play
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

MoDE (Mixture of Diffusion Experts) claims that occluded face recognition works better when a masked face is repainted several times by a diffusion model and the recognition scores of all repainted versions, together with the original occluded image, are combined by a learned identity gate in the decision space. The authors argue that this beats both recognizing the occluded image directly and averaging the repainted faces, because the gate can down-weight reconstructions whose random sampling drifted from the true identity. They position MoDE as a plug-and-play module that attaches to existing recognition backbones, and they report consistent gains on three public benchmarks and two self-collected real-world occlusion datasets. If the claim holds, masked and otherwise occluded face recognition can be improved without retraining the underlying recognition model.

What carries the argument

The central mechanism is the combination of a mask-conditioned diffusion repainting expert and an identity-gating network (ID-Gate). Each expert runs a Denoising Diffusion Probabilistic Model with a repainting procedure that uses resampling: at denoising step $t$ the known region $m\odot x^{\mathrm{known}}_{t-1}$ is sampled from the input's noised distribution while the unknown region $(1-m)\odot x^{\mathrm{unknown}}_{t-1}$ comes from the denoiser, and the two are spliced with $r$ resampling passes to smooth boundaries. The ID-Gate is a multi-input softmax network that receives the extracted feature vectors of the occluded original $x_0$ and $n$ repainted faces $x_1,\dots,x_n$ and outputs weights $w(x_i)$; the final prediction is the weighted sum of the individual similarity matrices, $S_X=\sum_{i=0}^n w(x_i)s(x_i)$. The gate's job is to decide, per probe, which reconstructions carry trustworthy identity information and which should be down-weighted.

What would settle it

Take a fixed set of occluded probe images with known ground-truth masks, run MoDE with the exact mask and with masks corrupted by increasing amounts of mislabeled pixels, and compare Top-1 accuracy; if accuracy falls back to the no-repaint baseline as soon as the mask is slightly wrong, the central claim depends on perfect mask availability, which the paper does not establish for real images.

Watch

Extended reading notes

Core claim

The paper's central claim is that a stochastic generator plus a learned fusion rule beats any single reconstruction: the occluded image $x_0$ and $n$ diffusion-repainted faces are treated as $n+1$ experts, a shared recognizer extracts an embedding from each, and the ID-Gate network outputs weights $w(x_i)$ that combine the per-expert similarity matrices into $S_X=\sum_{i=0}^n w(x_i)s(x_i)$, which becomes the recognition score. The authors report that this module raises Top-1 and Top-5 accuracy for every base recognizer on the synthetic mask benchmarks, and for most on the wild datasets, and that the gains stack with patch-wise EMD re-ranking. The underlying reasoning is that the diffusion model's sampling diversity supplies multiple plausible identities, and the gate learns which of those are trustworthy for a given probe.

Load-bearing premise

The method needs a correct map of which pixels are occluded for every test image, and the paper never says where that map comes from in practice.

Editorial extensions

If this is right

  • MoDE transfers across recognition backbones: every baseline tested on the synthetic occluded benchmarks shows higher Top-1 accuracy when the module is attached, and most improve on the real-world datasets as well.
  • The identity gate beats simple equal averaging of repainted faces: the ablation on Occ CelebA shows that averaging repainted faces helps, but weighting them with ID-Gate helps more.
  • Recognition accuracy keeps rising with the number of diffusion experts up to about four or five, after which the gain plateaus.
  • MoDE and patch-wise EMD re-ranking are complementary: combining both produces the largest gains, such as a 19.6-point Top-1 improvement on Occ MS1M in the EMD+MoDE row.

Reading between the lines

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

  • The largest practical gap the paper leaves open is mask acquisition: the repainting equations require a precise occlusion map $m$, and a learned mask estimator or a gate that consumes mask uncertainty would make the module end-to-end.
  • Because the gate acts on similarity vectors rather than raw pixels, the same decision-space gating could attach to other stochastic reconstruction pipelines, such as super-resolution or 3D face reconstruction, where several plausible outputs must be reconciled for a downstream identity decision.
  • The plateau at four to five experts suggests a compute-accuracy trade-off: an adaptive policy could stop adding experts once the gate's weights stabilize.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 7 minor

Summary. The manuscript proposes MoDE (Mixture of Diffusion Experts), a plug-and-play module for occluded face recognition. Given an occluded face image and a binary mask m, the method runs several diffusion-based inpainting experts to produce n repainted images, keeps the original occluded image as expert E0, extracts features from all n+1 images with a fixed face recognition backbone, and trains an ID-Gate network to assign weights to the corresponding similarity vectors before the final argmax classification. Experiments are reported on synthetically occluded MS1M, LFW, and CelebA, plus two self-collected real datasets OVF and WWCF, using ArcFace, FaceNet, CosFace, FFR-Net, and DeepFace-EMD as backbones or baselines. The main claimed contributions are the identity-gated fusion in decision space and the use of diffusion repainting for occluded face recognition.

Significance. If the central claim were established, MoDE would be a useful and practical contribution: decision-space fusion of multiple inpainting hypotheses is a sensible way to exploit generative diversity, and the plug-and-play design is attractive for real deployments. The two self-collected datasets and the systematic comparison with several face recognition models are also valuable. However, the manuscript's own tables contradict the headline claim that 'MoDE outperforms all other methods,' and a core test-time assumption (the availability of an accurate occlusion mask) is never addressed for real-world data. These issues must be resolved before the contribution is credible. The paper does not provide code or released datasets, which further limits verification, but the core idea remains potentially sound.

major comments (4)
  1. [Section IV-B1, Table I, Table II] The sentence 'MoDE outperforms all the competing methods' in Section IV-B1 and the Table I caption 'Our MoDE outperforms all other methods' are directly contradicted by the paper's own numbers. On Occ CelebA, the DeepFace-EMD row reports EMD Top-1 of 32.2 versus MoDE's 31.6; on OVF, EMD achieves 90.9 Top-1 versus MoDE's 87.5. In addition, on WWCF, the FaceNet row shows MoDE at 7.6 Top-1, below the FaceNet baseline of 8.6. Since MoDE is presented as a plug-and-play module, the fair comparison is against the same backbone, and in two settings the EMD re-ranking alone beats MoDE alone. The central claim should be revised to something like 'MoDE consistently improves its own backbone, and MoDE combined with DeepFace-EMD achieves the best results,' or the authors should provide a corrected comparison.
  2. [Section III-A, Eqs. (4)-(6)] The repainting step conditions on a binary mask m: Eq. (4) splices known pixels sampled via Eq. (5) with unknown pixels sampled via Eq. (6). The paper never explains how m is obtained at test time for real-world images or for the OVF and WWCF datasets. For synthetic datasets, the text says masks are created using facial keypoint detection, but real deployment requires a mask estimation procedure. If the mask is inaccurate, the known-region sampling in Eq. (5) will mix corrupted pixels into every diffusion step, degrading the reconstruction. The authors should specify the mask estimation method or explicitly state that m is assumed to be given by an external oracle.
  3. [Section IV-A] The training and testing protocol for Occ MS1M is under-specified. The text states that during training 500 faces in Occ MS1M are used as probe and 2500 faces in MS1M as gallery, and during testing 500 faces in Occ MS1M are randomly selected as probe with 2500 MS1M faces as gallery. There is no statement that the training and test probes or galleries are disjoint. If the same identities or even the same images are reused, the reported Occ MS1M results could reflect ID-Gate memorization rather than generalization. The authors must clarify the exact split and, ideally, evaluate on a held-out set of identities.
  4. [Section IV-B2] The discussion of the real-world results says 'MoDE performs better in dealing with face occlusion problems compared with most face recognition models.' This is too weak compared with the earlier 'outperforms all' claim, and it is also inconsistent with the FaceNet WWCF result where MoDE is worse than the baseline. The narrative should be aligned with the actual table entries, and the claims should be stated per backbone rather than as an unconditional superiority.
minor comments (7)
  1. [Section III-B, Eq. (8)] The notation is inconsistent: Eq. (7) defines W(X), the weight vector for the whole set X, while Eq. (8) uses w(x_i) for individual experts. Please clarify the relationship between the vector W(X) and the scalar weights w(x_i).
  2. [Section IV-A] The parameter 'jumpy size j' is listed in Implementation Details but is never defined or used in the equations. Define it in Section III-A or remove it from the text.
  3. [Section III-A] There is a typo: 'aj iterative resampling step' should be 'an iterative resampling step.'
  4. [References] Several reference entries contain 'n.d.' placeholders for publisher, volume, and page information, for example references [4], [6], [7], [35], and [42]. The bibliography should be completed before publication.
  5. [Section III-D] The text says 'the second Noisy Top-K Gating is discussed in the appendix,' but no appendix is present in the manuscript. Either include the appendix or remove the reference.
  6. [Section I and II] The contribution bullet 'our work is the first to introduce the diffusion model to occluded face recognition' is a strong novelty claim and is not supported by an exhaustive survey. Please soften the claim or provide a more complete literature comparison.
  7. [Throughout] The figures are referenced in the text (Figs. 2-7) but are not visible in the provided manuscript text; please ensure all figures are embedded and legible in the final submission.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: MoDE's reconstruction, gating, and fusion are trained on image and label data and evaluated on held-out benchmarks; no derivation reduces to its own inputs.

full rationale

No significant circularity is present. The paper's pipeline has three learned components: the DDPM-based repainting model, trained with Eq. (9) on image reconstruction targets; the face-recognition feature extractors, which are pretrained on MS1M and treated as fixed backbones; and the ID-Gate, trained with the cross-entropy loss in Eq. (11) on similarity-matrix classification labels. None of these training objectives assumes the final recognition claim. The diffusion experts are trained on images without identity labels, and the ID-Gate weights are fit on a training probe and then evaluated on held-out data, including Occ LFW, Occ CelebA, OVF, and WWCF. The gate's objective is to maximize classification accuracy, which is ordinary supervised training rather than a derivation that assumes its own conclusion. The mask-availability assumption in Eqs. (4)-(6) is a practical limitation, not a circular reduction, because it concerns test-time inputs rather than the derivation of the method. The in-domain Occ MS1M evaluation is less informative because the gate is trained on the same dataset domain, but the central claims are also supported by external benchmarks. The reviewer-flagged contradiction with DeepFace-EMD in Tables I and II is an internal-consistency and correctness issue, not a self-referential derivation. No load-bearing self-citation, imported uniqueness theorem, or ansatz-smuggling citation was found. The paper is self-contained against external benchmarks, so the circularity score is 0.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

No new physical entities are introduced. The central claims rest on trainable gate parameters, hand-chosen sampling hyperparameters, and domain assumptions about mask availability, identity preservation, and synthetic-to-real transfer.

free parameters (4)
  • ID-Gate weight matrix Wg = learned, not reported
    Trained with cross-entropy loss in Eq. (11); central to weighting the expert predictions in Eq. (7).
  • ID-Gate bias bg = learned, not reported
    Trained with the gate in Eq. (7).
  • Number of diffusion experts n = 4 or 5 (reported optimal range)
    Fig. 6 shows Top-1 and Acc stabilize at 4 to 5 experts; the exact n used for the main tables is not stated.
  • Resampling times r and jump length j = r=10, j=10
    Chosen by hand in Section IV-A to improve boundary quality; no ablation is provided for these values.
assumptions (4)
  • domain assumption A correct binary occlusion mask m is available at test time for every input image.
    Eqs. (4)-(6) split the image into known and unknown regions using m; no mask detection or annotation method is described for the real datasets.
  • domain assumption Diffusion repainting preserves the identity of the person while removing the occlusion.
    The entire expert ensemble depends on repainted faces being usable for recognition; Section III-A assumes this without a dedicated identity-preservation constraint.
  • domain assumption Synthetic masks generated by facial keypoint detection on MS1M, LFW, and CelebA are a valid proxy for real-world occlusions, so ID-Gate training transfers to the wild datasets.
    Section IV-A creates Occ datasets by adding mask textures; Section IV-B2 applies the model to OVF and WWCF with no retraining details.
  • standard math The DDPM equations of Ho et al. and the Repaint resampling procedure of Lugmayr et al. are correct and apply as background.
    Eqs. (1)-(3) and (9)-(10) import these published results without re-derivation; this is standard background rather than an ad hoc assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MoDE: Mixture of Diffusion Experts for Any Occluded Face Recognition." pith.science (2026). https://pith.science/paper/QAW6GMUJ

@misc{pith2026250504306,
  author       = {Pith},
  title        = {Pith review of: MoDE: Mixture of Diffusion Experts for Any Occluded Face Recognition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QAW6GMUJ}},
  note         = {Machine review of arXiv:2505.04306}
}
read the original abstract

With the continuous impact of epidemics, people have become accustomed to wearing masks. However, most current occluded face recognition (OFR) algorithms lack prior knowledge of occlusions, resulting in poor performance when dealing with occluded faces of varying types and severity in reality. Recognizing occluded faces is still a significant challenge, which greatly affects the convenience of people's daily lives. In this paper, we propose an identity-gated mixture of diffusion experts (MoDE) for OFR. Each diffusion-based generative expert estimates one possible complete image for occluded faces. Considering the random sampling process of the diffusion model, which introduces inevitable differences and variations between the inpainted faces and the real ones. To ensemble effective information from multi-reconstructed faces, we introduce an identity-gating network to evaluate the contribution of each reconstructed face to the identity and adaptively integrate the predictions in the decision space. Moreover, our MoDE is a plug-and-play module for most existing face recognition models. Extensive experiments on three public face datasets and two datasets in the wild validate our advanced performance for various occlusions in comparison with the competing methods.

Figures

Figures reproduced from arXiv: 2505.04306 by the authors.

Figure 1
Figure 1. To demonstrate the adaptability of our method for [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The framework of our MoDE. Firstly, MoDE reconstructs the occluded image and produces n repainted images. Then, [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Face Repainting. Repaint modifies the standard denois [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Visualization of similarity distributions with t-SNE [ [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 6
Figure 6. Figure 6: Hyperparameter analysis of the MoDE method with [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 5
Figure 5. Figure 5: Visualization of face repaint error. The first three [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 34 canonical work pages

  1. [1]

    Face recognition systems: A survey,

    Y . Kortli, M. Jridi, A. Al Falou, and M. Atri, “Face recognition systems: A survey,” Sensors, vol. 20, no. 2, p. 342, 2020

    2020

  2. [2]

    Deep learning face representa- tion by joint identification-verification,

    Y . Sun, Y . Chen, X. Wang, and X. Tang, “Deep learning face representa- tion by joint identification-verification,” Advances in neural information processing systems, vol. 27, p. n.d., 2014

    work page 2014

  3. [3]

    3d-aided dual- agent gans for unconstrained face recognition,

    J. Zhao, L. Xiong, J. Li, J. Xing, S. Yan, and J. Feng, “3d-aided dual- agent gans for unconstrained face recognition,” IEEE transactions on pattern analysis and machine intelligence , vol. 41, no. 10, pp. 2380– 2394, 2018

    work page 2018

  4. [4]

    Facenet: A unified embed- ding for face recognition and clustering,

    F. Schroff, D. Kalenichenko, and J. Philbin, “Facenet: A unified embed- ding for face recognition and clustering,” in Proceedings of the IEEE conference on computer vision and pattern recognition. n.d.: n.d., 2015, pp. 815–823

    work page 2015

  5. [5]

    Vggface2: A dataset for recognising faces across pose and age,

    Q. Cao, L. Shen, W. Xie, O. M. Parkhi, and A. Zisserman, “Vggface2: A dataset for recognising faces across pose and age,” in 2018 13th IEEE international conference on automatic face & gesture recognition (FG 2018), IEEE. n.d.: n.d., 2018, pp. 67–74

    work page 2018

  6. [6]

    Cosface: Large margin cosine loss for deep face recognition,

    H. Wang, Y . Wang, Z. Zhou, X. Ji, D. Gong, J. Zhou, Z. Li, and W. Liu, “Cosface: Large margin cosine loss for deep face recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition. n.d.: n.d., 2018, pp. 5265–5274

    work page 2018

  7. [7]

    Arcface: Additive angular margin loss for deep face recognition,

    J. Deng, J. Guo, N. Xue, and S. Zafeiriou, “Arcface: Additive angular margin loss for deep face recognition,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. n.d.: n.d., 2019, pp. 4690–4699

    work page 2019

  8. [8]

    Asymmetric joint learning for heterogeneous face recognition,

    B. Cao, N. Wang, X. Gao, and J. Li, “Asymmetric joint learning for heterogeneous face recognition,” in Proceedings of the AAAI Conference on Artificial Intelligence , vol. 32, no. 1, 2018

    work page 2018

Show all 45 references
  1. [9]

    A survey of face recognition techniques under occlusion,

    D. Zeng, R. Veldhuis, and L. Spreeuwers, “A survey of face recognition techniques under occlusion,” IET biometrics, vol. 10, no. 6, pp. 581–606, 2021

    2021

  2. [10]

    Face recognition with local binary patterns,

    T. Ahonen, A. Hadid, and M. Pietik ¨ainen, “Face recognition with local binary patterns,” in European conference on computer vision , Springer. n.d.: n.d., 2004, pp. 469–481

    2004

  3. [11]

    Face description with local binary patterns: Application to face recognition,

    T. Ahonen, A. Hadid, and M. Pietikainen, “Face description with local binary patterns: Application to face recognition,” IEEE transactions on pattern analysis and machine intelligence , vol. 28, no. 12, pp. 2037– 2041, 2006

    2006

  4. [12]

    A modular weighted sparse representation based on fisher discriminant and sparse residual for face recognition with occlusion,

    S. Zhao and Z.-p. Hu, “A modular weighted sparse representation based on fisher discriminant and sparse residual for face recognition with occlusion,” Information Processing Letters , vol. 115, no. 9, pp. 677– 683, 2015

    2015

  5. [13]

    Data augmentation-based joint learning for heterogeneous face recognition,

    B. Cao, N. Wang, J. Li, and X. Gao, “Data augmentation-based joint learning for heterogeneous face recognition,” IEEE transactions on neural networks and learning systems , vol. 30, no. 6, pp. 1731–1743, 2018

    2018

  6. [14]

    Ring loss: Convex feature nor- malization for face recognition,

    Y . Zheng, D. K. Pal, and M. Savvides, “Ring loss: Convex feature nor- malization for face recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition. n.d.: n.d., 2018, pp. 5089– 5097

    2018

  7. [15]

    A discriminative feature learning approach for deep face recognition,

    Y . Wen, K. Zhang, Z. Li, and Y . Qiao, “A discriminative feature learning approach for deep face recognition,” in Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11– 14, 2016, Proceedings, Part VII 14 , Springer. n.d.: n.d., 2016, pp. 499–515

    2016

  8. [16]

    Range loss for deep face recognition with long-tailed training data,

    X. Zhang, Z. Fang, Y . Wen, Z. Li, and Y . Qiao, “Range loss for deep face recognition with long-tailed training data,” in Proceedings of the IEEE International Conference on Computer Vision . n.d.: n.d., 2017, pp. 5409–5418

    2017

  9. [17]

    Large-margin softmax loss for convolutional neural networks,

    W. Liu, Y . Wen, Z. Yu, and M. Yang, “Large-margin softmax loss for convolutional neural networks,” arXiv preprint arXiv:1612.02295 , vol. n.d., no. n.d., p. n.d., 2016

    2016 arXiv

  10. [18]

    Deepface: Closing the gap to human-level performance in face verification,

    Y . Taigman, M. Yang, M. Ranzato, and L. Wolf, “Deepface: Closing the gap to human-level performance in face verification,” in Proceedings of the IEEE conference on computer vision and pattern recognition . n.d.: n.d., 2014, pp. 1701–1708

    2014

  11. [19]

    Multi-margin based decor- relation learning for heterogeneous face recognition,

    B. Cao, N. Wang, X. Gao, J. Li, and Z. Li, “Multi-margin based decor- relation learning for heterogeneous face recognition,” arXiv preprint arXiv:2005.11945, 2020

    2005 arXiv

  12. [20]

    Sphereface: Deep hypersphere embedding for face recognition,

    W. Liu, Y . Wen, Z. Yu, M. Li, B. Raj, and L. Song, “Sphereface: Deep hypersphere embedding for face recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition . n.d.: n.d., 2017, pp. 212–220

    2017

  13. [21]

    Face recognition by elastic bunch graph matching,

    L. Wiskott, N. Kr ¨uger, N. Kuiger, and C. V on Der Malsburg, “Face recognition by elastic bunch graph matching,” IEEE Transactions on pattern analysis and machine intelligence , vol. 19, no. 7, pp. 775–779, 1997

    1997

  14. [22]

    End2end occluded face recognition by masking corrupted features,

    H. Qiu, D. Gong, Z. Li, W. Liu, and D. Tao, “End2end occluded face recognition by masking corrupted features,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 44, no. 10, pp. 6939– 6952, 2021

    2021

  15. [23]

    Robust point set matching for partial face recognition,

    R. Weng, J. Lu, and Y .-P. Tan, “Robust point set matching for partial face recognition,” IEEE transactions on image processing, vol. 25, no. 3, pp. 1163–1176, 2016

    2016

  16. [24]

    Dynamic feature learning for partial face recognition,

    L. He, H. Li, Q. Zhang, and Z. Sun, “Dynamic feature learning for partial face recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition . n.d.: n.d., 2018, pp. 7054–7063

    2018

  17. [25]

    Occlusion robust face recognition based on mask learning with pairwise differential siamese network,

    L. Song, D. Gong, Z. Li, C. Liu, and W. Liu, “Occlusion robust face recognition based on mask learning with pairwise differential siamese network,” in Proceedings of the IEEE/CVF International Conference on Computer Vision. n.d.: n.d., 2019, pp. 773–782

    2019

  18. [26]

    Region filling and object removal by exemplar-based image inpainting,

    A. Criminisi, P. P ´erez, and K. Toyama, “Region filling and object removal by exemplar-based image inpainting,” IEEE Transactions on image processing, vol. 13, no. 9, pp. 1200–1212, 2004

    2004

  19. [27]

    Generative adversarial networks,

    I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y . Bengio, “Generative adversarial networks,” Communications of the ACM , vol. 63, no. 11, pp. 139–144, 2020

    2020

  20. [28]

    Attgan: Facial attribute editing by only changing what you want,

    Z. He, W. Zuo, M. Kan, S. Shan, and X. Chen, “Attgan: Facial attribute editing by only changing what you want,” IEEE transactions on image processing, vol. 28, no. 11, pp. 5464–5478, 2019

    2019

  21. [29]

    Occluded face recognition in the wild by identity-diversity inpainting,

    S. Ge, C. Li, S. Zhao, and D. Zeng, “Occluded face recognition in the wild by identity-diversity inpainting,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 30, no. 10, pp. 3387–3397, 2020

    2020

  22. [30]

    Face photo-sketch synthesis via full-scale identity supervision,

    B. Cao, N. Wang, J. Li, Q. Hu, and X. Gao, “Face photo-sketch synthesis via full-scale identity supervision,” Pattern Recognition , vol. 124, p. 108446, 2022

    2022

  23. [31]

    Autoencoder- based collaborative attention gan for multi-modal image synthesis,

    B. Cao, H. Cao, J. Liu, P. Zhu, C. Zhang, and Q. Hu, “Autoencoder- based collaborative attention gan for multi-modal image synthesis,” IEEE Transactions on Multimedia , vol. 26, pp. 995–1010, 2023

    2023

  24. [32]

    Autoencoder-driven multimodal collaborative learning for medical im- age synthesis,

    B. Cao, Z. Bi, Q. Hu, H. Zhang, N. Wang, X. Gao, and D. Shen, “Autoencoder-driven multimodal collaborative learning for medical im- age synthesis,” International Journal of Computer Vision, vol. 131, no. 8, pp. 1995–2014, 2023

    1995

  25. [33]

    Diffusion models beat gans on image synthesis,

    P. Dhariwal and A. Nichol, “Diffusion models beat gans on image synthesis,” Advances in Neural Information Processing Systems, vol. 34, pp. 8780–8794, 2021

    2021

  26. [34]

    Deep unsupervised learning using nonequilibrium thermodynamics,

    J. Sohl-Dickstein, E. Weiss, N. Maheswaranathan, and S. Ganguli, “Deep unsupervised learning using nonequilibrium thermodynamics,” in International Conference on Machine Learning , PMLR. n.d.: n.d., 2015, pp. 2256–2265

    2015

  27. [35]

    Repaint: Inpainting using denoising diffusion probabilistic models,

    A. Lugmayr, M. Danelljan, A. Romero, F. Yu, R. Timofte, and L. Van Gool, “Repaint: Inpainting using denoising diffusion probabilistic models,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . n.d.: n.d., 2022, pp. 11 461–11 471

    2022

  28. [36]

    Learning a prototype discriminator with rbf for multimodal image synthesis,

    Z. Bi, B. Cao, W. Zuo, and Q. Hu, “Learning a prototype discriminator with rbf for multimodal image synthesis,” IEEE Transactions on Image Processing, vol. 31, pp. 6664–6678, 2022

    2022

  29. [37]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” Advances in Neural Information Processing Systems , vol. 33, pp. 6840– 6851, 2020

    2020

  30. [38]

    Identity mappings in deep residual networks,

    K. He, X. Zhang, S. Ren, and J. Sun, “Identity mappings in deep residual networks,” in Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part IV 14. Springer, 2016, pp. 630–645

    2016

  31. [39]

    Hierarchical mixtures of experts and the em algorithm,

    M. I. Jordan and R. A. Jacobs, “Hierarchical mixtures of experts and the em algorithm,” Neural computation, vol. 6, no. 2, pp. 181–214, 1994

    1994

  32. [40]

    Outrageously large neural networks: The sparsely-gated mixture-of-experts layer,

    N. Shazeer, A. Mirhoseini, K. Maziarz, A. Davis, Q. Le, G. Hinton, and J. Dean, “Outrageously large neural networks: The sparsely-gated mixture-of-experts layer,” arXiv preprint arXiv:1701.06538 , vol. n.d., no. n.d., p. n.d., 2017

    2017 arXiv

  33. [41]

    A unified framework for masked and mask-free face recognition via feature rectification,

    S. Hao, C. Chen, Z. Chen, and K.-Y . K. Wong, “A unified framework for masked and mask-free face recognition via feature rectification,” in 2022 IEEE International Conference on Image Processing (ICIP), IEEE. n.d.: n.d., 2022, pp. 726–730

    2022

  34. [42]

    Deepface-emd: Re-ranking using patch-wise earth mover’s distance improves out-of-distribution face identification,

    H. Phan and A. Nguyen, “Deepface-emd: Re-ranking using patch-wise earth mover’s distance improves out-of-distribution face identification,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. n.d.: n.d., 2022, pp. 20 259–20 269

    2022

  35. [43]

    Labeled faces in the wild: A database forstudying face recognition in unconstrained environments,

    G. B. Huang, M. Mattar, T. Berg, and E. Learned-Miller, “Labeled faces in the wild: A database forstudying face recognition in unconstrained environments,” in Workshop on faces in’Real-Life’Images: detection, alignment, and recognition , no. n.d. n.d.: n.d., 2008, p. n.d

    2008

  36. [44]

    Deep learning face attributes in the wild,

    Z. Liu, P. Luo, X. Wang, and X. Tang, “Deep learning face attributes in the wild,” in Proceedings of the IEEE international conference on computer vision. n.d.: n.d., 2015, pp. 3730–3738

    2015

  37. [45]

    Visualizing data using t-sne

    L. Van der Maaten and G. Hinton, “Visualizing data using t-sne.” Journal of machine learning research , vol. 9, no. 11, p. n.d., 2008

    2008

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.