Pith. sign in

REVIEW 3 major objections 5 minor 56 references

Make a Face: Towards Arbitrary High Fidelity Face Manipulation

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · 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 a conditional variational auto-encoder, trained with only reconstruction and KL-divergence losses, can change a person's expression and head pose to an arbitrary target while keeping their identity and…

desk verdict A solid VAE alternative for face manipulation with convincing RaFD gains, but the headline CelebA SOTA number is a self-ablation, not an external baseline. read the letter →

arxiv 1908.07191 v1 pith:RC74DF5D submitted 2019-08-20 cs.CV

Shengju Qian , Kwan-Yee Lin , Wayne Wu , Yangxiaokang Liu , Quan Wang , Fumin Shen , Chen Qian , Ran He This is my paper · ORCID
classification cs.CV
keywords facemanipulationvariationalauto-encoderdisentangledrepresentationGaussianmixturepriorconditionalimagegenerationhigh-resolutionsynthesisexpressioneditingheadposerotation
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 claims that a conditional variational auto-encoder, trained with only reconstruction and KL-divergence losses, can change a person's expression and head pose to an arbitrary target while keeping their identity and high-resolution detail. The key move is to treat facial structure as a condition given by a target boundary map, and facial appearance as a latent code drawn from an additive Gaussian-mixture prior whose components are learned by clustering all boundary maps in the training set. On uncontrolled celebrity images this brings FID from 71.3 to 36.82 and Inception Score from 1.065 to 2.15, and in a human study the model's outputs fooled raters at a much higher rate than GAN baselines. If true, this is evidence that careful latent-space structure, rather than adversarial training alone, is the ingredient that enables high-fidelity conditional face synthesis.

What carries the argument

The workhorse is the additive focal prior: with K-means over the training boundary maps, each boundary gets a $K$-dimensional vector of cosine similarities to the cluster centers, and the prior on the appearance code is a single Gaussian whose mean is the weighted sum of those cluster means. That converts an intractable mixture prior into a closed-form KL divergence while keeping a multi-modal representation, which avoids the 'mean-face' collapse seen under extreme expressions or poses. Around this sit the geometry-guided disentanglement, where the encoded boundary map is concatenated with the appearance code before decoding, and two quality-aware design choices: sub-pixel upsampling to scatter checkerboard artifacts and weight normalization for faster, stabler training.

What would settle it

A direct test: take a fixed source identity and feed the model a target boundary whose facial contour comes from a different identity while keeping expression and pose fixed. If a face-recognition network measures a significant drop in identity similarity between source and output, the claim that appearance and structure are cleanly separated fails.

Watch

Extended reading notes

Core claim

The central claim is that arbitrary face manipulation can be achieved by decoupling an input face into a pose-invariant appearance code $z$ and a structure code $y$ obtained from a target boundary map, so that swapping the boundary map edits the expression or pose while the appearance code preserves identity. The paper's AF-VAE (Additive Focal Variational Auto-encoder) models the appearance prior as one Gaussian whose mean is the weighted sum of $K$ cluster means, which keeps the KL divergence closed-form while still giving a multi-modal latent space that can represent several facial characteristics at once. Two architecture choices, sub-pixel convolution and weight normalization, are shown to remove checkerboard artifacts and improve perceptual quality scores. The paper reports state-of-the-art FID and Inception Score on CelebA and higher fool rates than three GAN baselines in a human evaluation, and it shows that identity is preserved when the target boundary's facial contour is similar to the source while large contour differences shift identity.

Load-bearing premise

The central, load-bearing assumption is that an offline boundary map extracted from landmarks fully specifies facial structure without carrying identity; the paper's own failure cases show this assumption does not always hold when the target contour is very different from the source.

Editorial extensions

If this is right

  • If the central claim is correct, high-resolution face manipulation is achievable without paired training data or adversarial losses, using only a target boundary map as the user control.
  • The reported CelebA gains (FID 36.82 vs 71.3, IS 2.15 vs 1.065) imply that a VAE with a structurally richer prior can beat GAN-based methods on standard perceptual fidelity metrics.
  • The additive mixture prior should generalize to any conditional generation task where the conditioning signal can be clustered, since it leaves the KL term computable while expanding latent capacity.
  • The identity-preservation results indicate that facial identity is carried largely by facial contour and geometry, so extreme re-posing should be treated as a structure-transfer problem rather than an appearance-generation problem.
  • Human raters being fooled at a much higher rate than for GAN baselines suggests that current automatic metrics may understate the perceptual quality of non-adversarial generators.

Reading between the lines

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

  • Because the structure condition is just a clustered boundary map, the same additive-prior recipe could be dropped into other conditional image synthesis settings, such as full-body pose transfer or segmentation-guided scene editing, wherever the conditioning signal supports clustering.
  • A quantitative identity-preservation benchmark over held-out identities would settle how far the 'identity lives in structure' claim extends, since the paper's human study addresses realism but not identity retention directly.
  • The large gap between the model's 36% fool rate and the baselines' single-digit rates suggests that task-specific human evaluation may be a more informative comparison for face manipulation than generic generative-model metrics.
  • The paper's own failure cases point to a natural next step: enriching the boundary representation so that facial contour carries less identity-specific information, which would push arbitrary manipulation closer to full identity control.
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

3 major / 5 minor

Summary. The paper proposes Additive Focal Variational Auto-Encoder (AF-VAE) for high-resolution face manipulation. The method combines a geometry-guided disentanglement of appearance and structure using boundary maps, an additive focal prior built from K-means clustering over facial boundary maps, and two architecture-level choices (sub-pixel convolution and weight normalization) justified by human visual system arguments. The authors report comparisons with StarGAN, GANimation, and pix2pixHD on RaFD, ablations on CelebA, a mechanical Turk user study, interpolation results, and a discussion of failure cases. The central claim is that AF-VAE improves the state of the art in FID/IS on uncontrolled CelebA from 71.3/1.065 to 36.82/2.15 while preserving identity.

Significance. If the quantitative claims hold, the paper would demonstrate a competitive variational alternative to GAN-based face manipulation, with attractive properties such as stable training and multi-modal generation. The RaFD comparison in Table 1 and the user study in Table 2 provide concrete evidence that the proposed model is competitive with or better than three strong GAN baselines. The paper also includes a useful ablation study and openly discusses failure cases, which is a strength. However, the headline CelebA state-of-the-art claim is not supported as stated because the reported baseline 71.3/1.065 coincides with the paper's own ablation row, and the KL divergence derivation in Eq. (5) appears to be formally incorrect as written. These issues affect the manuscript's central quantitative and theoretical claims.

major comments (3)
  1. [Abstract and Section 1; Table 3] The claim that AF-VAE improves state-of-the-art FID/IS on uncontrolled CelebA from 71.3 and 1.065 to 36.82 and 2.15 is not supported by the evidence in the paper. Table 3 reports exactly 71.309 and 1.065 as the row 'Ours w/o PS,' which is an ablation of the proposed model rather than an external published method. No citation to an external algorithm is given for these baseline numbers. Please either supply a genuine external CelebA baseline with citation, or revise the abstract and introduction to describe this as a self-comparison with an ablated variant.
  2. [Section 3.2, Eq. (5)] The KL divergence in Eq. (5) is written as a scalar expression, but z, μφ, and the cluster means μk are latent vectors. For the stated q(z|x,y)=N(z|μφ,σφ^2 I) and prior p(z|y)=N(z|Σ wk μk, σ^2 I), the KL divergence contains a sum over the latent dimensions; e.g., for an isotropic posterior variance σφ^2 it should be d log(σ/σφ) + (d σφ^2 + ||μφ - Σ wk μk||^2)/(2σ^2) - d/2, up to the exact parametrization. The printed scalar formula omits the dimension factor and the norm. Please clarify whether the implementation uses the printed approximation or the correct vector form, and correct the derivation accordingly.
  3. [Section 4.3 and Figures 10(c), 11] The word 'arbitrary' in the title and abstract is stronger than what the paper's own experiments support. The failure cases in Fig. 10(c) and Fig. 11 show that when the target boundary has a substantially different facial contour from the source, the method changes identity rather than preserving it. Since the structure representation y is only an encoded boundary map, this is an expected limitation and is acknowledged in Sec. 4.3; nevertheless, the claims in the abstract and introduction should be tempered to reflect the identity-boundary constraint demonstrated by these experiments.
minor comments (5)
  1. [Section 3.2] The 'additive focal prior' in Eq. (4) is a single Gaussian with an additively combined mean, not a Gaussian mixture. The relationship between the Gaussian mixture prior in Eq. (3) and the actual training prior in Eq. (4) should be stated more precisely, since the ablation label 'Ours w/o GMM' is otherwise ambiguous.
  2. [Section 3.2] The focal weights w(b) are described as normalized cosine similarities to K-means cluster centers, but the exact normalization, whether the weights sum to one, and whether they are re-estimated during training are not specified. Please give the full definition and update procedure.
  3. [Section 4.1.2, Table 2] The AMT user study reports fool rate and TrueSkill, but the main text does not specify how many images per method were rated, how target boundary maps were selected, or whether raters saw paired source/target conditions. Please include these protocol details so the human evaluation can be assessed.
  4. [Section 2, reference [33]] The proposed additive Gaussian prior is closely related to the additive Gaussian encoding space of reference [45], and reference [33] is a prior workshop paper by the same group. The novelty of the proposed focal prior over both prior works should be discussed more explicitly.
  5. [Throughout] There are minor typographical inconsistencies, including 'AM-VAE' in the introduction where 'AF-VAE' is intended, and inconsistent use of φ and ϕ for encoder parameters in Eqs. (1) and (2).

Circularity Check

1 steps flagged · score 4.0 of 10

CelebA SOTA claim is self-referential: the abstract's 71.3/1.065 baseline is exactly the paper's own 'Ours w/o PS' ablation, not an external algorithm.

  1. other [Abstract; Table 3 (Sec. 4.2 Ablation Study)]
    "Abstract: 'The proposed AM-VAE improves the state-of-the-art Fréchet Inception Distance (FID) and Inception Score (IS) on uncontrolled CelebA from 71.3 and 1.065 to 36.82 and 2.15.' Table 3: 'Ours w/o PS 71.309 1.065' and 'Ours 36.820 2.152'."

    The claimed 'previous state-of-the-art' numbers on uncontrolled CelebA are never attributed to any published method; they are numerically identical to the paper's own 'Ours w/o PS' ablation row (71.309 vs 71.3; 1.065 vs 1.065). Thus the headline improvement is an internal difference between the full model and an ablated version of the same model, not a comparison against StarGAN/GANimation/pix2pixHD or any external baseline. The quantitative 'state-of-the-art' claim therefore reduces to a self-comparison: the baseline is the paper's own variant, so the improvement is not independent evidence of superiority.

full rationale

The core AF-VAE derivation is self-contained: the generative objective (Eqs. 1-2), the additive focal prior (Eqs. 3-4), and the closed-form KL (Eq. 5) are all derived in the paper, with the cluster centers obtained by K-means on training boundary maps and evaluated on held-out identities; FID/IS are computed on test-set generations. The additive-Gaussian motivation cites the authors' prior workshop paper [33] together with Wang et al. [45], but the formula and training objective are re-derived here, so this self-citation is contextual rather than load-bearing. The RaFD comparison (Table 1) and the AMT user study (Table 2) do use external baselines (pix2pixHD, StarGAN, GANimation). The paper's own Sec. 4.3 and Fig. 10-11 explicitly acknowledge that identity is not preserved when the target contour differs strongly from the source; this limits the 'arbitrary' claim but is a stated limitation, not a circular step. The one substantive circularity-adjacent defect is the CelebA SOTA baseline: the abstract's 'from 71.3 and 1.065' exactly matches the 'Ours w/o PS' ablation row in Table 3, with no external citation, so the headline quantitative claim rests on an internal ablation gap rather than an external state-of-the-art comparison. This is a baseline-equivalence issue in the central evaluation claim, while the method's derivation itself is not circular.

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

The method rests on fitted modeling choices: the number of focal clusters K=8, the cluster centers and variances from K-means on training boundary maps, and unspecified perceptual-loss weights. It also assumes boundary maps are a sufficient transferable geometry condition, that clusters are semantically meaningful, and that the scalar KL formula in Eq.5 applies to the latent vector. No invented entities are introduced.

free parameters (4)
  • Number of focal clusters K = 8
    Chosen by hand in Section 4 Implementation Details; no sensitivity analysis reported.
  • Cluster centers mu_k = Not reported (K-means centers of training boundary maps)
    Used as the additive prior mean in Eq.4; fitted to the training set.
  • Cluster variances sigma_k^2 = Not reported
    Define the prior covariance in Eq.4; values are not given in the main text.
  • Perceptual loss weights lambda_l = Not reported
    Used in the reconstruction loss in Section 3.3; values omitted.
assumptions (4)
  • domain assumption Boundary maps from pretrained landmark interpolation are a sufficient and transferable specification of facial structure for expression and pose manipulation.
    Used throughout Section 3.1; if a target boundary from another identity cannot be imposed without changing identity, the central manipulation claim fails. The paper's own Figure 10c shows identity change when contours differ.
  • domain assumption K-means clusters of training boundary maps correspond to semantic facial states such as laughing or side face.
    Stated in Section 3.2; the claimed multi-modal diversity of the additive prior depends on this semantic correspondence.
  • standard math Equation 5 is the correct KL divergence under the additive Gaussian prior for the latent space.
    Used to define the training loss; the printed formula appears to be a scalar expression while z is a vector, so the dimension factor is an unstated assumption.
  • domain assumption The human visual system weighs low-frequency and edge artifacts more heavily than high-frequency artifacts.
    Used in Section 3.3 to explain why sub-pixel convolution and weight normalization improve perceptual scores; cited from other papers, not measured here.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Make a Face: Towards Arbitrary High Fidelity Face Manipulation." pith.science (2026). https://pith.science/paper/RC74DF5D

@misc{pith2026190807191,
  author       = {Pith},
  title        = {Pith review of: Make a Face: Towards Arbitrary High Fidelity Face Manipulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RC74DF5D}},
  note         = {Machine review of arXiv:1908.07191}
}
read the original abstract

Recent studies have shown remarkable success in face manipulation task with the advance of GANs and VAEs paradigms, but the outputs are sometimes limited to low-resolution and lack of diversity. In this work, we propose Additive Focal Variational Auto-encoder (AF-VAE), a novel approach that can arbitrarily manipulate high-resolution face images using a simple yet effective model and only weak supervision of reconstruction and KL divergence losses. First, a novel additive Gaussian Mixture assumption is introduced with an unsupervised clustering mechanism in the structural latent space, which endows better disentanglement and boosts multi-modal representation with external memory. Second, to improve the perceptual quality of synthesized results, two simple strategies in architecture design are further tailored and discussed on the behavior of Human Visual System (HVS) for the first time, allowing for fine control over the model complexity and sample quality. Human opinion studies and new state-of-the-art Inception Score (IS) / Frechet Inception Distance (FID) demonstrate the superiority of our approach over existing algorithms, advancing both the fidelity and extremity of face manipulation task.

Figures

Figures reproduced from arXiv: 1908.07191 by the authors.

Figure 1
Figure 1. Face manipulation results on in-the-wild samples via transferring knowledge learned from the CelebA dataset. The first column shows input images and the remainders are images generated by AF-VAE with target expression/rotation boundary maps as the condition. Note that the model is fine-tuned with movie clip frames from YouTube of 256×256 resolution. All the generated poses are unseen before. Abstract Recent studies … view at source ↗
Figure 2
Figure 2. (a) Our framework (b) 2D and 3D projection of 5000 facial structure representations. Each color denotes a cluster, a more intuitive illustration of our approach is to map each cluster to a Gaussian prior, extending the capacity of C-VAE constrained by single prior. Eµ and Dϕ are further incorporated to potentially ensure enough structure information obtained from the prior (i.e., the decoder Dϕ). Thus, z is encourag… view at source ↗
Figure 3
Figure 3. Histograms w/wo pixel-shuffle (Better zoom in). and 43.6% boosts to IS and FID respectively. 4. Experiments Our framework provides a flexible way to manipulate an input face image into an arbitrary expression and pose un￾der the control of boundary maps. In this section, we show qualitative and quantitative comparisons with state-of-the￾art approaches in Sec 4.1. Then, we perform self-evaluation to analyze the key c… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Comparison with three state-of-the-art algorithms.(Zoom in for better details. Left three pictures are input faces, the right three lines are generated results of the three algorithms respectively. StarGAN [9] and GANimation [32] are best current approaches under large…
Figure 5
Figure 5. Figure 5: Face rotation results on 3D synthetic face dataset.There are 6 blocks, the upper left block represents 5 target boundary maps. For each blocks, the upper left face with red box is the input image and the rest 5 are synthesized results corresponding to 5 boundary maps i…
Figure 6
Figure 6. Figure 6: Face rotation results on MultiPIE [12] dataset. Performance metrics. For quantitative comparison, we evaluate three aspects, i.e. the realism, the perceptual qual￾ity and the diversity, for the synthesis results. For the measurement of perceptual quality and diversity,…
Figure 8
Figure 8. Figure 8: Ablation study on RafD, MultiPIE and CelebA datasets. Qualitative ablation. As shown in [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Interpolation results on CelebA. Upper left and lower right are two real images, each row and line represents linear In￾terpolation on latent appearance and structure vector, respectively. face and EmotionNet datasets, as shown in [PITH_FULL_IMAGE:figures/full_fig_p00…
Figure 11
Figure 11. Figure 11: Failure Cases. All four failure cases are selected from CelebA and EmotionNet dataset. We represent the source image at left, followed by manipulation result and its boundary reference. 5. Conclusion In this paper, we propose an additive focal variational auto-encoder…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 50 canonical work pages

  1. [1]

    CV AE-GAN: fine-grained image generation through asymmetric training

    Jianmin Bao, Dong Chen, Fang Wen, Houqiang Li, and Gang Hua. CV AE-GAN: fine-grained image generation through asymmetric training. In ICCV, 2017. 3

    work page 2017

  2. [2]

    Towards open-set identity preserving face synthesis

    Jianmin Bao, Dong Chen, Fang Wen, Houqiang Li, and Gang Hua. Towards open-set identity preserving face synthesis. In CVPR, 2018. 2, 7

    work page 2018

  3. [3]

    A morphable model for the synthesis of 3d faces

    V olker Blanz and Thomas Vetter. A morphable model for the synthesis of 3d faces. In Proceedings of the 26th an- nual conference on Computer graphics and interactive tech- niques, 1999. 2

    work page 1999

  4. [4]

    Large scale gan training for high fidelity natural image synthesis

    Andrew Brock, Jeff Donahue, and Karen Simonyan. Large scale gan training for high fidelity natural image synthesis. arXiv preprint arXiv:1809.11096, 2018. 3

    arXiv 2018

  5. [5]

    Facewarehouse: A 3d facial expression database for visual computing

    Chen Cao, Yanlin Weng, Shun Zhou, Yiying Tong, and Kun Zhou. Facewarehouse: A 3d facial expression database for visual computing. IEEE Trans. Vis. Comput. Graph. , 20(3):413–425, 2014. 8

    work page 2014

  6. [6]

    Photographic image syn- thesis with cascaded refinement networks

    Qifeng Chen and Vladlen Koltun. Photographic image syn- thesis with cascaded refinement networks. In ICCV, 2017. 5

    work page 2017

  7. [7]

    Facelet- bank for fast portrait manipulation

    Ying-Cong Chen, Huaijia Lin, Michelle Shu, Ruiyu Li, Xin Tao, Xiaoyong Shen, Yangang Ye, and Jiaya Jia. Facelet- bank for fast portrait manipulation. In CVPR, 2018. 2

    work page 2018

  8. [8]

    Zeyuan Chen, Shaoliang Nie, Tianfu Wu, and Christopher G. Healey. High resolution face completion with multiple con- trollable attributes via fully end-to-end progressive genera- tive adversarial networks. arXiv preprint arXiv:1801.07632,

    arXiv

Show all 56 references
  1. [9]

    Stargan: Unified genera- tive adversarial networks for multi-domain image-to-image translation

    Yunjey Choi, Minje Choi, Munyoung Kim, Jung-Woo Ha, Sunghun Kim, and Jaegul Choo. Stargan: Unified genera- tive adversarial networks for multi-domain image-to-image translation. In CVPR, 2018. 2, 5, 6, 7

    2018

  2. [10]

    Uv-gan: Adversarial facial uv map completion for pose-invariant face recognition

    Jiankang Deng, Shiyang Cheng, Niannan Xue, Yuxiang Zhou, and Stefanos Zafeiriou. Uv-gan: Adversarial facial uv map completion for pose-invariant face recognition. In CVPR, 2018. 8

    2018

  3. [11]

    A varia- tional u-net for conditional appearance and shape generation

    Patrick Esser, Ekaterina Sutter, and Bj ¨orn Ommer. A varia- tional u-net for conditional appearance and shape generation. In CVPR, 2018. 5

    2018

  4. [12]

    Multi-pie

    Ralph Gross, Iain Matthews, Jeffrey Cohn, Takeo Kanade, and Simon Baker. Multi-pie. Image and Vision Computing,

  5. [13]

    Improved training of wasserstein gans

    Ishaan Gulrajani, Faruk Ahmed, Martin Arjovsky, Vincent Dumoulin, and Aaron C Courville. Improved training of wasserstein gans. In NIPS, 2017. 5

    2017

  6. [14]

    Trueskill: a bayesian skill rating system

    Ralf Herbrich, Tom Minka, and Thore Graepel. Trueskill: a bayesian skill rating system. In NIPS, 2007. 6

    2007

  7. [15]

    Gans trained by a two time-scale update rule converge to a local nash equilib- rium

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule converge to a local nash equilib- rium. In NIPS, 2017. 2, 6

    2017

  8. [16]

    Pose-guided photorealistic face rotation

    Yibo Hu, Xiang Wu, Bing Yu, Ran He, and Zhenan Sun. Pose-guided photorealistic face rotation. In CVPR, 2018. 2, 6

    2018

  9. [17]

    Introvae: Introspective variational autoen- coders for photographic image synthesis

    Huaibo Huang, Zhihang Li, Ran He, Zhenan Sun, and Tieniu Tan. Introvae: Introspective variational autoen- coders for photographic image synthesis. arXiv preprint arXiv:1807.06358, 2018. 3

    2018 arXiv

  10. [18]

    Beyond face rotation: Global and local perception gan for photoreal- istic and identity preserving frontal view synthesis

    Rui Huang, Shu Zhang, Tianyu Li, and Ran He. Beyond face rotation: Global and local perception gan for photoreal- istic and identity preserving frontal view synthesis. InICCV,

  11. [19]

    Batch normalization: Accelerating deep network training by reducing internal co- variate shift

    Sergey Ioffe and Christian Szegedy. Batch normalization: Accelerating deep network training by reducing internal co- variate shift. arXiv preprint arXiv:1502.03167, 2015. 5

    2015 arXiv

  12. [20]

    Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, and Alexei A. Efros. Image-to-image translation with conditional adver- sarial networks. In CVPR, 2017. 3

    2017

  13. [21]

    Kingma and Max Welling

    Diederik P. Kingma and Max Welling. Auto-encoding vari- ational bayes. In ICLR, 2014. 2, 3

    2014

  14. [22]

    Gagan: Geometry-aware generative adversarial networks

    Jean Kossaifi, Linh Tran, Yannis Panagakis, and Maja Pantic. Gagan: Geometry-aware generative adversarial networks. In CVPR, 2018. 2

    2018

  15. [23]

    Pre- sentation and validation of the radboud faces database

    Oliver Langner, Ron Dotsch, Gijsbert Bijlstra, Daniel HJ Wigboldus, Skyler T Hawk, and AD Van Knippenberg. Pre- sentation and validation of the radboud faces database. Cog- nition and emotion, 2010. 5

    2010

  16. [24]

    Hallucinated-iqa: No- reference image quality assessment via adversarial learning

    Kwan-Yee Lin and Guanxiang Wang. Hallucinated-iqa: No- reference image quality assessment via adversarial learning. In CVPR, 2018. 5

    2018

  17. [25]

    Bagdanov

    Xialei Liu, Joost van de Weijer, and Andrew D. Bagdanov. Rankiqa: Learning from rankings for no-reference image quality assessment. In ICCV, 2017. 5

    2017

  18. [26]

    Deep learning face attributes in the wild

    Ziwei Liu, Ping Luo, Xiaogang Wang, and Xiaoou Tang. Deep learning face attributes in the wild. In ICCV, 2015. 2, 5

    2015

  19. [27]

    Attribute- guided face generation using conditional cyclegan

    Yongyi Lu, Yu-Wing Tai, and Chi-Keung Tang. Attribute- guided face generation using conditional cyclegan. InECCV,

  20. [28]

    Pose guided person image genera- tion

    Liqian Ma, Xu Jia, Qianru Sun, Bernt Schiele, Tinne Tuyte- laars, and Luc Van Gool. Pose guided person image genera- tion. In NIPS, 2017. 2

    2017

  21. [29]

    Disentangled per- son image generation

    Liqian Ma, Qianru Sun, Stamatios Georgoulis, Luc Van Gool, Bernt Schiele, and Mario Fritz. Disentangled per- son image generation. In CVPR, 2018. 2

    2018

  22. [30]

    Least squares generative adversarial networks

    Xudong Mao, Qing Li, Haoran Xie, Raymond YK Lau, Zhen Wang, and Stephen Paul Smolley. Least squares generative adversarial networks. In ICCV, 2017. 5

    2017

  23. [31]

    Blind predicting similar quality map for image quality assessment

    Da Pan, Ping Shi, Ming Hou, Zefeng Ying, Sizhe Fu, and Yuan Zhang. Blind predicting similar quality map for image quality assessment. In CVPR, 2018. 5

    2018

  24. [32]

    Martinez, Al- berto Sanfeliu, and Francesc Moreno-Noguer

    Albert Pumarola, Antonio Agudo, Aleix M. Martinez, Al- berto Sanfeliu, and Francesc Moreno-Noguer. Ganimation: Anatomically-aware facial animation from a single image. In ECCV, 2018. 2, 5, 6, 7

    2018

  25. [33]

    Extending the capacity of cvae for face synthesis and modeling

    Shengju Qian, Wayne Wu, Yangxiaokang Liu, Beier Zhu, and Fumin Shen. Extending the capacity of cvae for face synthesis and modeling. In NeurIPS Workshops, 2018. 3

    2018

  26. [34]

    Geometry-contrastive generative adversarial network for facial expression synthesis

    Fengchun Qiao, Naiming Yao, Zirui Jiao, Zhihao Li, Hui Chen, and Hongan Wang. Geometry-contrastive generative adversarial network for facial expression synthesis. arXiv preprint arXiv:1802.01822, 2018. 2

    2018 arXiv

  27. [35]

    Improved techniques for training gans

    Tim Salimans, Ian Goodfellow, Wojciech Zaremba, Vicki Cheung, Alec Radford, and Xi Chen. Improved techniques for training gans. In NIPS, 2016. 6

    2016

  28. [36]

    Weight normaliza- tion: A simple reparameterization to accelerate training of deep neural networks

    Tim Salimans and Diederik P Kingma. Weight normaliza- tion: A simple reparameterization to accelerate training of deep neural networks. In NIPS, 2016. 5

    2016

  29. [37]

    Faceid-gan: Learning a symmetry three-player gan for identity-preserving face synthesis

    Yujun Shen, Ping Luo, Junjie Yan, Xiaogang Wang, and Xi- aoou Tang. Faceid-gan: Learning a symmetry three-player gan for identity-preserving face synthesis. In CVPR, 2018. 2, 7

    2018

  30. [38]

    Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network

    Wenzhe Shi, Jose Caballero, Ferenc Husz ´ar, Johannes Totz, Andrew P Aitken, Rob Bishop, Daniel Rueckert, and Zehan Wang. Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network. In CVPR, 2016. 5

    2016

  31. [39]

    Deform- ing autoencoders: Unsupervised disentangling of shape and appearance

    Zhixin Shu, Mihir Sahasrabudhe, Riza Alp G ¨uler, Dimitris Samaras, Nikos Paragios, and Iasonas Kokkinos. Deform- ing autoencoders: Unsupervised disentangling of shape and appearance. In ECCV, 2018. 3

    2018

  32. [40]

    Neural face editing with intrinsic image disentangling

    Zhixin Shu, Ersin Yumer, Sunil Hadap, Kalyan Sunkavalli, Eli Shechtman, and Dimitris Samaras. Neural face editing with intrinsic image disentangling. In CVPR, 2017. 3

    2017

  33. [41]

    Learning structured output representation using deep conditional gen- erative models

    Kihyuk Sohn, Honglak Lee, and Xinchen Yan. Learning structured output representation using deep conditional gen- erative models. In NIPS, 2015. 2

    2015

  34. [42]

    Face2face: Real-time face capture and reenactment of rgb videos

    Justus Thies, Michael Zollhofer, Marc Stamminger, Chris- tian Theobalt, and Matthias Nießner. Face2face: Real-time face capture and reenactment of rgb videos. In CVPR, 2016. 2

    2016

  35. [43]

    Nonlinear 3d face morphable model

    Luan Tran and Xiaoming Liu. Nonlinear 3d face morphable model. In CVPR, 2018. 2

    2018

  36. [44]

    On learning 3d face mor- phable model from in-the-wild images

    Luan Tran and Xiaoming Liu. On learning 3d face mor- phable model from in-the-wild images. arXiv preprint arXiv:1808.09560, 2018. 2

    2018 arXiv

  37. [45]

    Diverse and accurate image description using a variational auto-encoder with an additive gaussian encoding space

    Liwei Wang, Alexander Schwing, and Svetlana Lazebnik. Diverse and accurate image description using a variational auto-encoder with an additive gaussian encoding space. In NIPS, 2017. 3

    2017

  38. [46]

    Video-to- video synthesis

    Ting-Chun Wang, Ming-Yu Liu, Jun-Yan Zhu, Guilin Liu, Andrew Tao, Jan Kautz, and Bryan Catanzaro. Video-to- video synthesis. In NIPS, 2018. 2

    2018

  39. [47]

    High-resolution image syn- thesis and semantic manipulation with conditional gans

    Ting-Chun Wang, Ming-Yu Liu, Jun-Yan Zhu, Andrew Tao, Jan Kautz, and Bryan Catanzaro. High-resolution image syn- thesis and semantic manipulation with conditional gans. In CVPR, 2018. 3, 5, 7

    2018

  40. [48]

    Every smile is unique: Landmark-guided diverse smile generation

    Wei Wang, Xavier Alameda-Pineda, Dan Xu, Pascal Fua, Elisa Ricci, and Nicu Sebe. Every smile is unique: Landmark-guided diverse smile generation. In CVPR, 2018. 2

    2018

  41. [49]

    Self- supervised learning of a facial attribute embedding from video

    Olivia Wiles, A Koepke, and Andrew Zisserman. Self- supervised learning of a facial attribute embedding from video. In BMVC, 2018. 2

    2018

  42. [50]

    X2face: A network for controlling face generation using im- ages, audio, and pose codes

    Olivia Wiles, A Sophia Koepke, and Andrew Zisserman. X2face: A network for controlling face generation using im- ages, audio, and pose codes. In ECCV, 2018. 2

    2018

  43. [51]

    Transgaga: Geometry-aware unsuper- vised image-to-image translation

    Wayne Wu, Kaidi Cao, Cheng Li, Chen Qian, and Chen Change Loy. Transgaga: Geometry-aware unsuper- vised image-to-image translation. In CVPR, 2019. 2

    2019

  44. [52]

    Look at boundary: A boundary-aware face alignment algorithm

    Wayne Wu, Chen Qian, Shuo Yang, Quan Wang, Yici Cai, and Qiang Zhou. Look at boundary: A boundary-aware face alignment algorithm. In CVPR, 2018. 3

    2018

  45. [53]

    Reenactgan: Learning to reenact faces via boundary transfer

    Wayne Wu, Yunxuan Zhang, Cheng Li, Chen Qian, and Chen Change Loy. Reenactgan: Learning to reenact faces via boundary transfer. In ECCV, 2018. 2

    2018

  46. [54]

    Pose guided human video generation

    Ceyuan Yang, Zhe Wang, Xinge Zhu, Chen Huang, Jianping Shi, and Dahua Lin. Pose guided human video generation. In ECCV, 2018. 2

    2018

  47. [55]

    Towards large-pose face frontalization in the wild

    Xi Yin, Xiang Yu, Kihyuk Sohn, Xiaoming Liu, and Man- mohan Chandraker. Towards large-pose face frontalization in the wild. In ICCV, 2017. 2

    2017

  48. [56]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In CVPR, 2018. 5

    2018

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.