Dual-reference Age Synthesis

and Jun He; BingZhang Hu; Ling Shao; Yuan Zhou; Yu Guan

REVIEW 5 major objections 6 minor 44 references

Dual-reference Age Synthesis

T0 review · 5 major / 6 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 Facial age can be copied from one photo onto another person's identity, no age labels needed.

desk verdict A genuinely new conditioning signal for face aging—reference-image age features—with a plausible framework and decent ablations, but the age feature space is under-validated and the closest baseline is missing. read the letter →

arxiv 1908.02671 v4 pith:O36A2BXT submitted 2019-08-07 cs.CV

Yuan Zhou , Bingzhang Hu , and Jun He , Yu Guan , Ling Shao This is my paper · ORCID

classification cs.CV

keywords agesynthesisdualreferencesoftinformationgenerativeadversarialnetworkfaceagingrejuvenationidentitypreservation

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

Age synthesis usually asks a model to turn one face into the same person at a given number of years. This paper tries to establish a different task: turn an identity image into a face that looks the same age as a second, unrelated image. The motivation is that a photo carries 'soft' age information—rounder cheeks, wrinkles, skin texture—that a number or a word like 'old' cannot capture, and different people of the same age can look very different. The authors claim that with two feature agents and a GAN, a generator can swap the apparent age of the age reference onto the identity reference, and their experiments on two benchmark face datasets report stronger age preservation than a numeric-conditioned baseline while keeping identity.

What carries the argument

The load-bearing mechanism is the pair of feature agents and their two preservation losses. The identity agent is an encoder $E_I$ coupled to a discriminator $D_I$ that pushes identity codes toward a uniform prior, so identity features of different people separate from one another; the age agent $E_A$ is a truncated pretrained apparent-age network whose final 50-dimensional output is used as the age code. The generator receives the concatenation of the two codes, and the two Euclidean losses $\|E_I(\tilde{I}^n_i)-E_I(I^m_i)\|_2$ and $\|E_A(\tilde{I}^n_i)-E_A(I^n_j)\|_2$ are what force the output to inherit identity from one image and age from the other. The discriminator $D$ and the reconstruction loss keep the output photo-realistic and aligned with the identity reference's pixel structure.

What would settle it

Take a fixed identity image and two age reference images that human raters judge to be the same apparent age but that differ sharply in pose or expression, and generate outputs with each. If an independent age estimator or human raters judge the two outputs to differ in apparent age by about as much as outputs produced from references known to be different ages, then the age feature is carrying identity or pose information and the central transfer assumption fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that facial age can be manipulated by an image reference instead of a scalar: the generator $G$ takes an identity feature $E_I(I^m_i)$ from the identity reference and an age feature $E_A(I^n_j)$ from the age reference and produces a face $\tilde{I}^n_i$ that satisfies both $E_I(\tilde{I}^n_i)\approx E_I(I^m_i)$ and $E_A(\tilde{I}^n_i)\approx E_A(I^n_j)$. The age agent is adapted from a pretrained apparent-age network by removing its last layer and adding two fully connected layers (1024 and 50 units), so the 50-dimensional output acts as a compact 'soft age' code. Training minimizes a hybrid loss of five terms: a pixel reconstruction loss, an adversarial loss on the image, an adversarial uniform-prior loss on identity features, an identity preservation loss, and the age preservation loss of Equation (1). Because the age reference is just another image sampled at random from the same dataset, the training needs no age annotations and no same-person cross-age pairs. The reported consequence is that DRAS synthesizes the identity at whatever apparent age the reference image displays, with higher age-preservation accuracy than a numeric-conditioned age model and better feature-space identity separation than a reconstruction-only model.

Load-bearing premise

The load-bearing premise is that the 50-dimensional age feature really captures apparent age and nothing else, so two faces whose features are close in Euclidean distance will look the same age even when their identities, poses, and expressions differ.

Editorial extensions

If this is right

Age synthesis becomes feasible on datasets that have neither age labels nor pairs of the same person at different ages, since the age reference image supplies supervision through the age feature.
Swapping the age reference image changes the apparent age of the output, so a user can specify 'looks like this person's age' rather than a number, including subtle within-group differences.
The framework separates identity and age into different feature spaces during training, which is what lets it age the same identity consistently across ten age groups.
On the reported benchmark comparisons, the model's age-preservation accuracy exceeds the numeric-conditioned baseline while its identity verification confidence stays above the acceptance threshold.
Numeric age conditioning becomes a special case: choose an age reference image that looks the target age and the same generation mechanism applies.

Reading between the lines

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

The 'no age labels' claim applies to DRAS training, not to the age feature itself, which is inherited from a network pretrained on labeled apparent-age data; a stricter test of the paradigm would start the age agent from an unsupervised face embedding.
The dual-reference formulation should transfer to other attributes that resist numeric labels, such as expression intensity, fatigue, or apparent health, by replacing the age agent with a feature extractor for that attribute.
A library-based interface is a natural extension: instead of a slider, a user selects an exemplar face, and the model interpolates or extrapolates between identity and age references.
Because only a single 50-dimensional vector represents age, using distributional divergence between feature distributions (a direction the authors themselves flag) may improve robustness when several reference images define the target age.

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.

Desk Editor's Note

read the letter

Colleague,

The one thing to know: this paper replaces the hard numeric age label with a 50-dimensional age feature extracted from a second face image, and that is a real change of task. The dual-reference setup—keep identity from image A, adopt apparent age from image B—is new as far as I can tell, and the framework is assembled from solid parts: CAAE-style identity encoder, DEX-initialized age encoder, two discriminators. The ablations show both preservation losses earn their keep, and using Face++ verification plus an external AlexNet age classifier is better evidence than most age-synthesis papers offer.

The weakness cluster is centered on the age feature space. The method's entire mechanism is the Euclidean distance in E_A (Eq. 1), but the paper never shows that this 50-d space is age-only or even age-monotonic after training. DEX gives a good prior, but the two new fully-connected layers are trained with reconstruction and same-image age preservation, which can let identity and pose leak into E_A. The AlexNet evaluation shows coarse age-group accuracy, but it does not validate Euclidean distance as a perceptual age metric. Second, the closest prior work, Dual Conditional GANs [41], is mentioned once and never compared. Third, there are no trivial baselines—for instance, feeding CAAE the age label of the reference image would test whether the extra image is actually helping. Fourth, age-preservation accuracy lacks error bars, and IPCGAN's five age groups are not comparable with DRAS's ten. No code is released.

None of these, by themselves, sink the paper. The reported age-group accuracy is real evidence, and the method plausibly works. But the 'soft age' claim is stronger than the experiments: all reference images in the evaluation carry age-group labels, so we never see the case where soft conditioning matters, i.e., a reference with ambiguous perceived age.

Who should read it: people working on face aging or conditional image translation who want a label-free reference-based condition. It deserves a serious referee. I would send it to review, asking for an analysis of the age feature space (age-ordered t-SNE, a linear probe, an identity-leakage test) and the missing comparisons.

Referee Report

5 major / 6 minor

Summary. The paper proposes dual-reference age synthesis (DRAS), a GAN-based framework that takes two facial images as input: one identity reference and one age reference. The age agent maps the age reference to a 50-dimensional feature, the identity agent maps the identity reference to a latent identity code, and the generator synthesizes an output that is meant to share identity with the first image and age with the second. The model is trained with reconstruction, adversarial, identity-preservation, and age-preservation losses, and is evaluated on UTKFace and CACD against CAAE and IPCGAN. The authors claim that DRAS provides a more flexible 'soft' age control than numeric conditioning and requires no paired data or age annotations.

Significance. If the central mechanism works, the paper introduces a genuinely useful task formulation: age synthesis conditioned on a reference image rather than a scalar age. This could simplify annotation requirements and better capture the perceived-age variability that a single number misses. The framework itself is clearly described, and the idea of using an externally pretrained age estimator (DEX) as the age feature extractor is reasonable. The paper also includes ablation studies and quantitative comparisons, and it reports machine-checkable experimental numbers. However, the central claim depends on an unvalidated assumption about the age feature space, and the experimental protocol omits the closest prior work and trivial baselines, so the significance of the empirical results is currently not established.

major comments (5)

[§3.2, Eq. (1)] The age preservation loss minimizes the Euclidean distance between EA of the generated image and EA of the age reference. The entire age-transfer mechanism therefore rests on the assumption that the 50-dimensional EA space is a faithful, transferable representation of perceived age and is largely invariant to identity, pose, and expression. The paper provides no direct evidence for this. The two newly added fully-connected layers are trained from DEX initialization using L_age, L_rec, and the adversarial losses without any age labels or monotonicity constraint, so nothing prevents EA from collapsing or from encoding generator-specific artifacts. The paper's own age-preservation evaluation uses a separate AlexNet classifier fine-tuned on age labels, not the internal EA metric, so it does not validate that Euclidean distance in EA corresponds to perceived age. Please provide evidence for age-monotonicity or cross-identity age transfer in the EA space, for example by evaluating age retrieval accuracy using EA features, comparing fixed DEX features against the fine-tuned ones, or showing that matching EA produces the target age in a controlled experiment.
[§4.2 and §4.3.3] The closest prior work, Dual Conditional GANs [41], is mentioned in Section 4.2 as a method that DRAS differs from, but it is never included in the experimental comparisons. Given that [41] also uses two conditional inputs for face aging, the statement that DRAS achieves the best performance among compared methods is incomplete without a comparison to this baseline. If [41] is excluded for a concrete reason, that reason should be stated explicitly; otherwise the comparison should be added.
[§4.3.2, Tables 4, 9, 10] The age-preservation accuracies are reported without error bars, confidence intervals, or significance tests. Since the accuracies are often near saturation (e.g., 99.93% vs 99.80% for group 0#), the reported differences may not be statistically meaningful. Please report standard deviations over multiple training runs or apply a significance test, and state how many test images underlie each accuracy value.
[§4.3.4, Tables 5-8] The identity-preservation comparison with IPCGAN is confounded. In Section 4.3.3 and Figure 12 the authors note that IPCGAN's synthesized images look almost the same as their identity reference images, with little visible aging effect. It is therefore expected that IPCGAN achieves the highest verification confidence, since it trivially preserves identity by failing at the age-transfer task. A meaningful identity-preservation comparison must control for the magnitude of the applied age change, for example by measuring identity preservation only on images where the age reference is correctly matched.
[§4.2 and §4.3, Fig. 12] No trivial baseline is reported. A simple baseline that outputs the identity reference unchanged, or a version of DRAS trained without the age reference, would clarify whether the age reference contributes any measurable effect beyond the reconstruction objective. Reporting identity and age preservation for such a baseline would strengthen the claim that the dual-reference mechanism itself is responsible for the observed performance.

minor comments (6)

[§3.1, Eq. (3)] In Eq. (3), the notation 'EI∼pdata (I)' is confusing; it should be 'I ∼ pdata(I)' in the data term. Also, the two discriminators DI and D are both used in the text with subscripts, but later the implementation section refers to them without consistent notation.
[§4.3.1] The sentence 'T-Distributed Stochastic Neighbor Embedding(t-SNE) [42,43]depictsthesimilaritiesofidentityfeatures, which can be used to visualize the disentangled their disentangled representations' is grammatically broken and should be rewritten.
[§4.1] The phrase 'images with rank higher than five are low quality' refers to CACD's rank annotation, but the notion of rank is not defined for a reader who is not already familiar with the CACD dataset. A brief definition or reference would help.
[§4.3.2] The sentence 'which explain the reconstruction loss plays an important role in identity perservation' contains typos and should read 'which explains that the reconstruction loss plays an important role in identity preservation.'
[Eq. (6)] In Eq. (6) the loss weights are displayed with placeholder glyphs in the submitted PDF; they should be typeset as ordinary Greek symbols (e.g., λ_adv, λ_id, λ_age) to be legible.
[§3.2] The term 'soft age information' is used informally. Since it is a central conceptual contribution, consider giving a precise definition of what 'soft' means in terms of the feature-space representation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the age representation is anchored to an external pretrained DEX model and age transfer is evaluated with an independent classifier.

full rationale

The paper's training objective defines age transfer as minimizing ||EA(I_n_j) - EA(G(EI(I_m_i), EA(I_n_j)))|| (Eq. 1), but EA is initialized from the externally pretrained DEX age estimator [37,38] and its last fully connected layers are only lightly adapted without age labels before being frozen for generator training (Sec. 4.2, Stage 1/2). The target age is therefore not defined solely by the model's own outputs; DEX provides an independent, label-trained anchoring. The age-preservation evaluation uses a separate AlexNet fine-tuned on UTKFace/CACD age labels rather than the trained EA, so the reported 'same age group' accuracies are not the same quantity as the training loss. Identity preservation is likewise evaluated with Face++ and t-SNE on EI. The framework borrows GAN/AAE architectures and a uniform identity prior from prior work [10] by citation, but those are standard external building blocks, not self-citations, and the central dual-reference formulation is not reduced to them. No fitted parameter is renamed as a prediction, and no uniqueness theorem or author-chain citation is used to force the design. The main limitation flagged by the skeptic, that DEX features may not be age-only or age-monotonic, is a correctness/robustness assumption, not a circularity, because the paper checks age transfer externally.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The method relies on the pretrained DEX age representation, the CAAE identity architecture and uniform prior, and a two-stage training strategy that uses self-reconstruction without paired data. The main free parameters are the loss weights and the age feature dimension, neither of which is ablated.

free parameters (3)

lambda_id (identity preservation weight) = 1e-3
Hand-chosen in Eq. (6); no sensitivity analysis provided.
lambda_age (age preservation weight) = 1e-2
Hand-chosen in Eq. (6); no sensitivity analysis provided.
age feature dimension = 50
Two FC layers (1024 and 50) in the age agent; dimension chosen without ablation.

assumptions (4)

domain assumption A face image lies on two low-dimensional manifolds, identity and age, that change smoothly and are sufficiently disentangled to allow independent feature extraction.
Section 3.1 states the manifold assumption; the entire architecture relies on independent identity and age features.
domain assumption DEX (pretrained for age estimation) provides a feature space where Euclidean distance reflects age similarity, and adapting only the last two FC layers preserves this property.
Section 3.2 introduces the age agent; no validation that L2 distance in this space is perceptually meaningful.
domain assumption The identity feature is assumed to follow a uniform distribution, following CAAE [10], which regularizes the identity manifold.
Section 3.3, Eq. (3), follows prior work without re-validation.
ad hoc to paper Training with a reconstruction loss using the same image as both references (Stage 1) is sufficient to learn meaningful identity and age features without paired data.
Section 4.2 describes the joint-training strategy; the success of this self-supervision is assumed, not proven.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Dual-reference Age Synthesis." pith.science (2026). https://pith.science/paper/O36A2BXT

@misc{pith2026190802671,
  author       = {Pith},
  title        = {Pith review of: Dual-reference Age Synthesis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O36A2BXT}},
  note         = {Machine review of arXiv:1908.02671}
}

read the original abstract

Age synthesis methods typically take a single image as input and use a specific number to control the age of the generated image. In this paper, we propose a novel framework taking two images as inputs, named dual-reference age synthesis (DRAS), which approaches the task differently; instead of using "hard" age information, i.e. a fixed number, our model determines the target age in a "soft" way, by employing a second reference image. Specifically, the proposed framework consists of an identity agent, an age agent and a generative adversarial network. It takes two images as input - an identity reference and an age reference - and outputs a new image that shares corresponding features with each. Experimental results on two benchmark datasets (UTKFace and CACD) demonstrate the appealing performance and flexibility of the proposed framework.

Figures

Figures reproduced from arXiv: 1908.02671 by the authors.

**Figure 1.** Comparison between a conventional age synthesis framework and our proposed dual-reference age synthesis framework. (a) Conventional GAN-based age synthesis framework: a one-hot-vector as an age feature is transformed from the target age, then an identity reference image and the one-hot-vector are fed into an generator to synthesize a facial image. (b) Dual-reference Age Synthesis framework: the framework synthesizes… view at source ↗

**Figure 2.** Demonstration of our age synthesis results (images with black dotted boxes are the original ones) [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** The framework of the proposed age synthesis method. The two raw reference images are taken as input to synthesize images which share the same identity information with the identity reference images and having the same age information with the age reference images. Different people of the same age often have different age appearances. Therefore, rather than providing a fixed numerical “hard” age, it is more reasonabl… view at source ↗

Figures from the paper (13 more)

**Figure 4.** Figure 4: Original data distributions of CACD and UTKFace. Âă Âă Âă Âă Âă (a) 3 (b) 9 (c) 13 (d) 17 (e) 23 (f) 32 (g) 45 (h) 54 (i) 69 (j) 79 Âă [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

**Figure 5.** Figure 5: Ten age reference images. The numbers under each age reference image are the age annotations. Âă 4.3. Experimental Performance and Analysis In this section, we first investigate the performance of our model, then select two baselines, CAAE [10] and IPCGAN [11], for co…

**Figure 6.** Figure 6: Visualization of identity features. (a) Age ranges from 14 to 26 years old; (b)Age ranges from 20 to 33 years old; (c) Age ranges from 43 to 58 years old. images of those overlapping points. It can be seen that some of overlapping are caused by similar makeups, e.g. 6 …

**Figure 7.** Figure 7: Visualization of identity features in different age ranges. (a) 0, 9 and 10 are from three different age ranges; (b) 0 and 2 are from the same age range, and 7 is from another age range. In these two experiments, most of the identity features of different people fall i…

**Figure 8.** Figure 8: Effects of the identity and age preservation functions. From top to bottom, facial images are generated by 1, 2, 3 and 4, respectively. In [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]

**Figure 9.** Figure 9: Effects of identity preservation function. From top to bottom: identity feature visualizations obtained under 1, 2, 3 and 4, respectively. ∗ is the synthesized image for identity reference image . (a) 14-26 years old; (b) 20-33 years old; (c) 43-58 years old. confidenc…

**Figure 10.** Figure 10: Effects of identity preservation loss. From top to bottom: visualizations of identity features obtained under 1, 2, 3 and 4, respectively. ∗ is the synthesized image for identity reference image . (a) 1, 7 and 8 are from three different age ranges; (b) 0 and 3 are fro…

**Figure 11.** Figure 11: Quantified identity consistency of different age groups with 4 models. Y-axis shows the average verification confidence between age group with age groups [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]

**Figure 12.** Figure 12: Some synthesized faces for UTKFace and CACD. Each dotted box denotes images of the same person. The first column on the left in each dotted box is the identity reference image for DRAS (the input images for CAAE and IPCGAN). The images generated for the five age group…

**Figure 13.** Figure 13: Reconstruction results of the CAAE, IPCGAN and DRAS. Images in the left dotted box are from UTKFace and images in the right dotted box are from CACD. Images with red boxes are different from their ground truth or blurry in local facial regions. Furthermore, facial ima…

**Figure 14.** Figure 14: Synthesized faces with different identity reference images and age reference images. The first row shows the age reference images, while the first column on the left shows the identity reference images. The first, second and third row in each dotted box show the resul…

**Figure 15.** Figure 15: Identity preservation comparison of DRAS, IPCGAN and CAAE. From left to right: they are identity feature visualizations of DRAS, IPCGAN and CAAE, respectively. ∗ is the synthesized image of identity reference image . Points with dotted circles are far away from their …

**Figure 16.** Figure 16: Identity preservation performance of DRAS, IPCGAN and CAAE for different age ranges. From left to right: they are identity feature visualizations of DRAS, IPCGAN and CAAE, respectively. ∗ is the synthesized image of identity reference image . (a) 1, 7 and 8 are from t…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 40 canonical work pages

[41]

J. Song, J. Zhang, L. Gao, X. Liu, H. T. Shen, Dual conditional gans for face aging and rejuvenation., in: IJCAI, 2018, pp. 899–905

work page 2018
[1]

Goodfellow, J

I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, Y. Bengio, Generative adversarial nets, in: Advances in neural information processing systems, 2014, pp. 2672–2680

work page 2014
[2]

Ricanek Jr, E

K. Ricanek Jr, E. Boone, E. Patterson, Craniofacial aging impacts on the eigenface face biometric, Computer Science 1 (2006) 3

work page 2006
[3]

Y.Fu,G.Guo,T.S.Huang,Agesynthesisandestimationviafaces: Asurvey,IEEEtransactionsonpatternanalysisandmachineintelligence 32 (11) (2010) 1955–1976

work page 2010
[4]

L. S. Mark, J. B. Pittenger, H. Hines, C. Carello, R. E. Shaw, J. T. Todd, Wrinkling and head shape as coordinated sources of age-level information, Perception & Psychophysics 27 (2) (1980) 117–124

work page 1980
[5]

A.J.O’toole,T.Vetter,H.Volz,E.M.Salter,Three-dimensionalcaricaturesofhumanheads: distinctivenessandtheperceptionoffacialage, Perception 26 (6) (1997) 719–732

work page 1997
[6]

A. J. O’Toole, T. Price, T. Vetter, J. C. Bartlett, V. Blanz, 3d shape and 2d surface textures of human faces: the role of â/uni0102/uni0132averagesâ/uni0102/uni0130 in attractiveness and age, Image and Vision Computing 18 (1) (1999) 9–19

work page 1999
[7]

J.Suo,S.-C.Zhu,S.Shan,X.Chen,Acompositionalanddynamicmodelforfaceaging,IEEETransactionsonPatternAnalysisandMachine Intelligence 32 (3) (2009) 385–401

work page 2009

Show all 44 references

[8]

Tazoe, H

Y. Tazoe, H. Gohara, A. Maejima, S. Morishima, Facial aging simulator considering geometry and patch-tiled texture, in: ACM SIGGRAPH 2012 Posters, 2012, pp. 1–1

2012
[9]

2378–2386

W.Wang,Z.Cui,Y.Yan,J.Feng,S.Yan,X.Shu,N.Sebe,Recurrentfaceaging,in: ProceedingsoftheIEEEConferenceonComputerVision and Pattern Recognition, 2016, pp. 2378–2386

2016
[10]

Zhang, Y

Z. Zhang, Y. Song, H. Qi, Age progression/regression by conditional adversarial autoencoder, in: The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Vol. 2, 2017

2017
[11]

Z. Wang, X. Tang, W. Luo, S. Gao, Face aging with identity-preserved conditional generative adversarial networks, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 7939–7947

2018
[12]

Tiddeman, M

B. Tiddeman, M. Burt, D. Perrett, Prototyping and transforming facial textures for perception research, IEEE computer graphics and applica- tions 21 (5) (2001) 42–50

2001
[13]

P. J. Benson, D. I. Perrett, Extracting prototypical facial images from exemplars, Perception 22 (3) (1993) 257–262

1993
[14]

Kemelmacher-Shlizerman, S

I. Kemelmacher-Shlizerman, S. Suwajanakorn, S. M. Seitz, Illumination-aware age progression, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2014, pp. 3334–3341

2014
[15]

Z. Liu, Z. Zhang, Y. Shan, Image-based surface detail transfer, IEEE Computer Graphics and Applications 24 (3) (2004) 30–35

2004
[16]

H. Yang, D. Huang, Y. Wang, A. K. Jain, Learning face age progression: A pyramid architecture of gans, arXiv preprint arXiv:1711.10352 (2017)

2017 arXiv
[17]

Antipov, M

G. Antipov, M. Baccouche, J.-L. Dugelay, Face aging with conditional generative adversarial networks, arXiv preprint arXiv:1702.01983 (2017)

2017 arXiv
[18]

Lanitis, C

A. Lanitis, C. J. Taylor, T. F. Cootes, Toward automatic simulation of aging eﬀects on face images, IEEE Transactions on pattern Analysis and machine Intelligence 24 (4) (2002) 442–455

2002
[19]

Mukaida, H

S. Mukaida, H. Ando, Extraction and manipulation of wrinkles and spots for facial image synthesis, in: Sixth IEEE International Conference on Automatic Face and Gesture Recognition, 2004. Proceedings., IEEE, 2004, pp. 749–754

2004
[20]

M. R. Gandhi, A method for automatic synthesis of aged human facial images, Ph.D. thesis, Citeseer (2004)

2004
[21]

Ramanathan, R

N. Ramanathan, R. Chellappa, Modeling age progression in young faces, in: 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’06), Vol. 1, IEEE, 2006, pp. 387–394

2006
[22]

Y. Fu, N. Zheng, M-face: An appearance-based photorealistic model for multiple facial attributes rendering, IEEE Transactions on Circuits and Systems for Video technology 16 (7) (2006) 830–842

2006
[23]

J.Suo,X.Chen,S.Shan,W.Gao,Q.Dai,Aconcatenationalgraphevolutionagingmodel,IEEEtransactionsonpatternanalysisandmachine intelligence 34 (11) (2012) 2083–2096

2012
[24]

P. Kaur, H. Zhang, K. J. Dana, Photo-realistic facial texture transfer, arXiv preprint arXiv:1706.04306 (2017)

2017 arXiv
[25]

X. Shu, J. Tang, H. Lai, L. Liu, S. Yan, Personalized age progression with aging dictionary, in: Proceedings of the IEEE international conference on computer vision, 2015, pp. 3970–3978

2015
[26]

X. Shu, J. Tang, H. Lai, Z. Niu, S. Yan, Kinship-guided age progression, Pattern Recognition 59 (2016) 156–167

2016
[27]

A. M. Bukar, H. Ugail, Facial age synthesis using sparse partial least squares (the case of ben needham), Journal of forensic sciences 62 (5) (2017) 1205–1212. Yuan Zhou et al.:Preprint submitted to Elsevier Page 18 of 19 Dual Reference Age Synthesis

2017
[28]

A. M. Bukar, H. Ugail, N. Hussain, On facial age progression based on modiﬁed active appearance models with face texture, in: Advances in computational intelligence systems, Springer, 2017, pp. 465–479

2017
[29]

T. F. Cootes, G. J. Edwards, C. J. Taylor, Active appearance models, IEEE Transactions on pattern analysis and machine intelligence 23 (6) (2001) 681–685

2001
[30]

Mirza, S

M. Mirza, S. Osindero, Conditional generative adversarial nets, arXiv preprint arXiv:1411.1784 (2014)

2014 arXiv
[31]

Makhzani, J

A. Makhzani, J. Shlens, N. Jaitly, I. Goodfellow, B. Frey, Adversarial autoencoders, arXiv preprint arXiv:1511.05644 (2015)

2015 arXiv
[32]

5967–5976

G.Lample,N.Zeghidour,N.Usunier,A.Bordes,L.Denoyer,etal.,Fadernetworks: Manipulatingimagesbyslidingattributes,in: Advances in Neural Information Processing Systems, 2017, pp. 5967–5976

2017
[33]

S.Liu, Y.Sun, D.Zhu, R.Bao, W.Wang, X.Shu, S.Yan, Faceagingwithcontextualgenerativeadversarialnets, in: Proceedingsofthe2017 ACM on Multimedia Conference, ACM, 2017, pp. 82–90

2017
[34]

H. Ding, K. Sricharan, R. Chellappa, Exprgan: Facial expression editing with controllable expression intensity, arXiv preprint arXiv:1709.03842 (2017)

2017 arXiv
[35]

Y. Choi, M. Choi, M. Kim, J.-W. Ha, S. Kim, J. Choo, Stargan: Uniﬁed generative adversarial networks for multi-domain image-to-image translation, in: The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018

2018
[36]

J. Bao, D. Chen, F. Wen, H. Li, G. Hua, Towards open-set identity preserving face synthesis, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 6713–6722

2018
[37]

R.Rothe,R.Timofte,L.V.Gool,Dex: Deepexpectationofapparentagefromasingleimage,in: IEEEInternationalConferenceonComputer Vision Workshops (ICCVW), 2015

2015
[38]

Rothe, R

R. Rothe, R. Timofte, L. V. Gool, Deep expectation of real and apparent age from a single image without facial landmarks, International Journal of Computer Vision (IJCV) (July 2016)

2016
[39]

J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, L. Fei-Fei, Imagenet: A large-scale hierarchical image database, in: 2009 IEEE conference on computer vision and pattern recognition, Ieee, 2009, pp. 248–255

2009
[40]

Chen, C.-S

B.-C. Chen, C.-S. Chen, W. H. Hsu, Cross-age reference coding for age-invariant face recognition and retrieval, in: European conference on computer vision, Springer, 2014, pp. 768–783

2014
[42]

L. v. d. Maaten, G. Hinton, Visualizing data using t-sne, Journal of machine learning research 9 (Nov) (2008) 2579–2605

2008
[43]

Van der Maaten, G

L. Van der Maaten, G. Hinton, Visualizing non-metric similarities in multiple maps, Machine learning 87 (1) (2012) 33–55

2012
[44]

Yuan Zhou et al.:Preprint submitted to Elsevier Page 19 of 19

Megvii, Face++ face comparing api, acessed December 2019 athttps://www.faceplusplus.com/face-comparing/. Yuan Zhou et al.:Preprint submitted to Elsevier Page 19 of 19

2019

Pith tools

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

[1] [41]

J. Song, J. Zhang, L. Gao, X. Liu, H. T. Shen, Dual conditional gans for face aging and rejuvenation., in: IJCAI, 2018, pp. 899–905

work page 2018

[2] [1]

Goodfellow, J

I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, Y. Bengio, Generative adversarial nets, in: Advances in neural information processing systems, 2014, pp. 2672–2680

work page 2014

[3] [2]

Ricanek Jr, E

K. Ricanek Jr, E. Boone, E. Patterson, Craniofacial aging impacts on the eigenface face biometric, Computer Science 1 (2006) 3

work page 2006

[4] [3]

Y.Fu,G.Guo,T.S.Huang,Agesynthesisandestimationviafaces: Asurvey,IEEEtransactionsonpatternanalysisandmachineintelligence 32 (11) (2010) 1955–1976

work page 2010

[5] [4]

L. S. Mark, J. B. Pittenger, H. Hines, C. Carello, R. E. Shaw, J. T. Todd, Wrinkling and head shape as coordinated sources of age-level information, Perception & Psychophysics 27 (2) (1980) 117–124

work page 1980

[6] [5]

A.J.O’toole,T.Vetter,H.Volz,E.M.Salter,Three-dimensionalcaricaturesofhumanheads: distinctivenessandtheperceptionoffacialage, Perception 26 (6) (1997) 719–732

work page 1997

[7] [6]

A. J. O’Toole, T. Price, T. Vetter, J. C. Bartlett, V. Blanz, 3d shape and 2d surface textures of human faces: the role of â/uni0102/uni0132averagesâ/uni0102/uni0130 in attractiveness and age, Image and Vision Computing 18 (1) (1999) 9–19

work page 1999

[8] [7]

J.Suo,S.-C.Zhu,S.Shan,X.Chen,Acompositionalanddynamicmodelforfaceaging,IEEETransactionsonPatternAnalysisandMachine Intelligence 32 (3) (2009) 385–401

work page 2009