Pith. sign in

REVIEW 3 major objections 6 minor 58 references

AI-driven Remote Facial Skin Hydration and TEWL Assessment from Selfie Images: A Systematic Solution

T0 review · 3 major / 6 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read Skin hydration and water loss can be estimated from selfie images

desk verdict A genuinely new dataset for selfie-based SH/TEWL, but the missing per-position mean baseline leaves the central claim unproven—still worth refereeing. read the letter →

arxiv 2509.06282 v1 pith:VTQNLJCB submitted 2025-09-08 cs.CV

Cecelia Soh , Rizhao Cai , Monalisha Paul , Dennis Sng , Alex Kot This is my paper
classification cs.CV
keywords skinhydrationestimationtransepidermalwaterlossselfieimageanalysisvisiontransformeradaptersimbalancedregressioncontrastivelearningbarrierfunctionfacialassessment
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

Dermatologists measure skin barrier function with point instruments—a Corneometer for stratum corneum hydration and a VapoMeter for trans-epidermal water loss—but these devices are expensive and clinic-bound. This paper claims that both quantities can instead be estimated from ordinary RGB selfie images, making full-face skin-barrier assessment as simple as taking a photo. To establish this, the authors built a dataset of 336 panelists measured at 37 standardized facial anchors and paired with smartphone and controlled VISIA images, and trained Skin-PA ViT, a frozen Vision Transformer augmented with texture and position adapters plus a symmetry-based contrastive regularizer. On held-out test patches, the model reaches R2 = 0.264 for TEWL and R2 = 0.205 for SH from Selfie images, outperforming standard CNN and ViT baselines, with the clearest gains on rare value ranges. If the claim holds, remote, low-cost skin-barrier monitoring—including heatmaps of where the barrier is weak—becomes feasible for general users.

What carries the argument

The load-bearing object is the Skin-Prior Adaptive Vision Transformer (Skin-PA ViT), a regression head built on a frozen ImageNet-pretrained ViT-B. Instead of tuning the backbone, the method trains three small components: a Prior Texture Module (a small CNN) whose output is turned into prompt tokens by Texture Adapters, and Position Adapters that convert the facial anchor ID (1 of 37) into prompt tokens prepended to the input sequence of every Transformer layer. The prompt tokens let each layer use both image texture and knowledge of where on the face the patch sits. The input patch is concatenated with a band-pass-filtered version of itself in the Fourier domain, which keeps mid-range spati

What would settle it

A position-only control: train Skin-PA ViT with the 37-position one-hot ID but a blank or averaged image (or otherwise ablate image information). If it reaches R2 near 0.264 for TEWL or 0.205 for SH on the Selfie test set, the central remote-assessment claim—that skin appearance is being read—would fail.

Watch

Extended reading notes

Core claim

The paper's central claim is that hydration (SH) and trans-epidermal water loss (TEWL), two point-measured markers of skin barrier function, are readable from small RGB patches of facial skin. The authors collected paired instrument measurements at 37 facial anchors and facial images from 336 panelists, then trained a regression model that predicts the two values from each cropped patch. Their model, Skin-PA ViT, keeps an ImageNet-pretrained Vision Transformer frozen and learns only lightweight adapters that inject high-frequency texture tokens and a one-hot facial-position token into every Transformer layer; a band-pass filtered version of the patch is used to highlight wrinkles and pores,

Load-bearing premise

The load-bearing premise is that a skin patch's appearance, not its labeled position on the face, is what carries the SH/TEWL signal; the model is given both, so the claim stands or falls on whether image texture is genuinely predictive once position is controlled.

Editorial extensions

If this is right

  • If the method generalizes beyond the collected population, an ordinary smartphone photo can produce full-face SH/TEWL heatmaps, removing the need for a clinic visit for routine barrier-function checks.
  • The position-adapter design shows that injecting explicit anatomical region information helps a small-data regression model, suggesting a recipe for other per-region physiological estimates from images.
  • The symmetry-based contrastive loss and lighting augmentation both reduce error on rare value ranges, pointing to a general strategy for imbalanced regression where labels are distributed unevenly across spatial regions.
  • The data-collection and anchor-estimation pipeline provides a reproducible protocol for building future skin-health image datasets with point-instrument labels.
  • Because SH prediction trails TEWL, the paper's own framing implies that adding non-RGB cues (for example infrared or other modalities) would be the natural next step for hydration.

Reading between the lines

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

  • The paper leaves implicit that a position-only control—training the same architecture with the 1-of-37 position ID but with image content ablated—would quantify how much of the R2 is regional prior rather than skin appearance; that experiment is a direct test a reader could run.
  • Because the dataset is drawn from 336 Chinese panelists with a skin-tone-based train/test split, the method's cross-ethnicity generalizability is untested; a natural extension is to evaluate on panels spanning broader ITA values and device types.
  • The anchor estimator's average error of 0.77–0.78 sticker radii means training labels are spatially noisy; one testable extension is to train with label-smoothing or multiple random crops per anchor and measure whether accuracy improves.
  • A consumer version would need to handle occlusions, extreme poses, and makeup; since the model is evaluated on multi-angle, multi-lighting images, a natural stress test is to probe those real-world distortions systematically.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper introduces a complete pipeline for estimating facial skin hydration (SH) and trans-epidermal water loss (TEWL) from smartphone and VISIA facial images. Data were collected from 336 Chinese panelists at 37 facial anchors using a Corneometer and a VapoMeter; anchors are localized in non-sticker images by a DAD-3DHeads + PointNet pipeline. The proposed Skin-PA ViT uses a frozen ImageNet-pretrained ViT with texture adapters, a prior texture module with frequency filtering, position adapters fed by 1-of-37 anchor IDs, lighting augmentation, and a facial-symmetry contrastive loss. On subject-disjoint test splits, the method reports R2 = 0.264 (TEWL) and 0.205 (SH) on Selfie images and R2 = 0.348 and 0.263 on VISIA images, outperforming several CNN and ViT baselines. Ablations attribute the gains incrementally to the proposed components.

Significance. If the reported predictive power comes from skin appearance rather than from positional priors, this would be a useful step toward accessible, image-based skin-barrier assessment. Strengths of the manuscript include a subject-disjoint train/test split stratified by skin tone, evaluation on two capture modalities, multiple strong baselines, and an ablation study that is reported transparently. The work is also honest about dataset limitations and the weaker performance on selfie images. However, the central claim that SH and TEWL are estimable from RGB patches is not yet established: the position-ID input is a major confound, and the reported anchor-estimation error loosens the image-label coupling. The missing per-position mean-only baseline is therefore load-bearing and must be addressed before the image-driven claim is convincing.

major comments (3)
  1. [§3.3, §4.7, Table 2] The Position Adapters take the 1-of-37 anchor ID as a one-hot input and prepend a position prompt at every transformer layer. Table 2 shows the largest single R2 jump when these adapters are added (Config B to C: Selfie TEWL 0.210→0.253, SH 0.109→0.190; VISIA TEWL 0.311→0.320, SH 0.200→0.231). Figure 8 shows that the per-position mean SH/TEWL values vary substantially across the face. A trivial predictor that outputs the training-set mean for each anchor ID could plausibly reproduce much of this gain. Because no such per-position mean baseline is reported, the reported R2 values are uninterpretable as evidence that the model reads skin appearance rather than memorizing regional averages. Please add a per-position mean-only baseline (and ideally a position-ID-only model with no image input) for both datasets and for the shot-stratified MAE columns. This is necessary to support the central
  2. [§4.4] The anchor estimator reports average error rates of 0.77 and 0.78 sticker radii for Selfie and VISIA images, and the paper states these are acceptable because the error is within the sticker radius. However, the SH/TEWL ground-truth labels are point measurements taken at the sticker centroids, while the test-time patches are cropped around the predicted anchors. An offset of ~0.78 radii can place the cropped patch mostly outside the region that was physically measured, weakening the image-label alignment. The paper does not quantify the effect of this misalignment on the reported R2. Please report the evaluation using ground-truth sticker centroids on the test set (or a sticker-jitter sensitivity analysis) to show how much performance degrades when anchor error is removed, and discuss the implications for the reliability of the learned image-label mapping.
  3. [§4.2, Table 1] All regression metrics are computed over cropped patches, but patches are highly correlated within a subject: the same subject appears in multiple views, lighting conditions, and nearby facial anchors. The differences between the proposed method and the strongest baselines are small in several cases (e.g., Selfie TEWL R2 0.264 vs. 0.236 for Swin-B; VISIA TEWL R2 0.348 vs. 0.318 for ViT-B). Without subject-level clustering, confidence intervals, or a paired test, it is unclear whether these differences are statistically meaningful. Please report cluster-robust standard errors or subject-level bootstrap confidence intervals for the headline R2 and MAE numbers.
minor comments (6)
  1. [§3.2 vs. §4.1] The paper defines M=37 anchor points per facial image, but §4.1 states that 19 skin patches are cropped from each facial image. This inconsistency affects the dataset-size calculation and the meaning of the position-ID input. Please clarify which anchors are visible or used per image/angle and correct the notation.
  2. [§3.3] The Position Adapter input is described as a one-hot vector of anchor IDs, but the relationship between the 37 global anchor IDs and the 19 patches per image is not specified. At test time, is the position ID the global facial anchor ID or an image-relative index? Please define the encoding precisely.
  3. [§3.5, Eq. (10)] The objective function uses Lcon in Eq. (10) while Eq. (9) defines ℓcon; please unify the notation. Also clarify the construction of the negative set Z in Eq. (9): are negatives drawn from other positions, other subjects, and/or other images in the batch? This affects the interpretation of the contrastive loss.
  4. [Table 2] The text says the final Config E achieves superior performance, but for VISIA TEWL the MAE (All) increases from 2.16 (Config D) to 2.19 (Config E). Please qualify the claim or discuss why the symmetric contrastive loss trades overall MAE for better few-shot MAE.
  5. [§4.6, Fig. 15] The lighting generalization results are presented only as a figure with average R2 improvements. Please report the numeric values, including the without-augmentation baseline for each leave-one-lighting-out direction, so the gains are quantitatively assessable.
  6. [Throughout] Minor language issues: 'par with' in §4.6, 'it's' in the Acknowledgment, and the duplicated caption text in Fig. 2. These do not affect the technical content.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: SH/TEWL predictions are empirical regressions from image patches to device-measured labels, evaluated on held-out subjects.

full rationale

The paper's central claims are empirical: a model is trained on skin patches with SH/TEWL labels obtained from Corneometer and VapoMeter measurements and evaluated on held-out subjects. The reported R2 values are genuine out-of-sample prediction results, not quantities derived by construction from the inputs. The label is a physical device measurement, the input is an image patch plus a position ID, and no equation in the paper defines the target in terms of the input or vice versa. The self-citations (e.g., refs. [35-40] on adapters for face anti-spoofing) are background related work and are not load-bearing evidence for the SH/TEWL assessment claim. The anchor-estimation error (0.77-0.78 sticker radii) is label noise rather than circularity. The Position Adapters feeding anchor ID into each transformer layer is a potential confound, because the model could partly memorize per-position averages (as the paper itself shows in Fig. 8), but this is an experimental-design/soundness concern, not a definitional or self-citational circularity. A per-position mean-only baseline would strengthen the interpretation, but its absence does not make the derivation circular.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the domain assumptions that visible skin appearance encodes hydration and barrier function, that symmetric face regions share similar values, that the 37-point device measurements are reliable labels despite imperfect anchor alignment, and that the PointNet anchor estimator generalizes to non-sticker images. The crop radius, frequency-filter thresholds, and contrastive temperature are tuned hyperparameters; the paper does not report all values.

free parameters (4)
  • Patch crop radius r = 70 px (Selfie), 170 px (VISIA)
    Chosen from average sticker radius in each imaging modality (Section 3.2); affects what texture the model sees.
  • Frequency filtering thresholds rho_l, rho_h = band-pass 0.36% to 5.76% of frequency spectrum
    Selected by comparing low-pass, band-pass, and high-pass variants (Section 4.7, Fig. 17); a tuned hyperparameter.
  • Contrastive temperature tau = not reported
    Hyperparameter in the symmetry contrastive loss (Eq. 9); value not stated in the paper.
  • Lighting augmentation factor m range = uniform random from unspecified range in [0,2]
    Magnitude of enhancement blending in Eq. 8; distribution not specified.
assumptions (5)
  • domain assumption SH and TEWL are visually manifested in RGB facial skin appearance (texture, color, wrinkles) so that regression from images is possible.
    Core premise of the entire approach; motivated by refs [3-6] in Sections 1-2 but not verified independently.
  • domain assumption Symmetric facial regions have similar SH/TEWL distributions, so contrastive alignment of symmetric patch features is valid.
    Used to define the symmetry contrastive loss (Section 3.5, Eq. 9-10); supported by average heatmaps (Fig. 8) but individual deviations are common.
  • domain assumption Point measurements at 37 stickers are reliable ground truth after 20-minute equilibration, and anchor errors within one sticker radius are acceptable label noise.
    Data collection protocol (Section 3.1) and anchor tolerance (Section 4.4).
  • domain assumption DAD-3DHeads landmarks and the trained PointNet transfer to non-sticker images.
    Anchor estimation pipeline (Section 3.2) assumes the landmark-to-anchor mapping is consistent across sticker and non-sticker conditions.
  • domain assumption ImageNet-pretrained ViT features are a useful starting point for skin patches.
    Initialization choice (Section 3.3) relies on transfer from natural images to skin texture.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AI-driven Remote Facial Skin Hydration and TEWL Assessment from Selfie Images: A Systematic Solution." pith.science (2026). https://pith.science/paper/VTQNLJCB

@misc{pith2026250906282,
  author       = {Pith},
  title        = {Pith review of: AI-driven Remote Facial Skin Hydration and TEWL Assessment from Selfie Images: A Systematic Solution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VTQNLJCB}},
  note         = {Machine review of arXiv:2509.06282}
}
read the original abstract

Skin health and disease resistance are closely linked to the skin barrier function, which protects against environmental factors and water loss. Two key physiological indicators can quantitatively represent this barrier function: skin hydration (SH) and trans-epidermal water loss (TEWL). Measurement of SH and TEWL is valuable for the public to monitor skin conditions regularly, diagnose dermatological issues, and personalize their skincare regimens. However, these measurements are not easily accessible to general users unless they visit a dermatology clinic with specialized instruments. To tackle this problem, we propose a systematic solution to estimate SH and TEWL from selfie facial images remotely with smartphones. Our solution encompasses multiple stages, including SH/TEWL data collection, data preprocessing, and formulating a novel Skin-Prior Adaptive Vision Transformer model for SH/TEWL regression. Through experiments, we identified the annotation imbalance of the SH/TEWL data and proposed a symmetric-based contrastive regularization to reduce the model bias due to the imbalance effectively. This work is the first study to explore skin assessment from selfie facial images without physical measurements. It bridges the gap between computer vision and skin care research, enabling AI-driven accessible skin analysis for broader real-world applications.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

58 extracted references · 47 canonical work pages

  1. [1]

    Syn- chronized in vivo measurements of skin hydration and trans-epidermal water loss. exploring their mutual influences,

    E. Caberlotto, C. Cornillon, S. Njikeu, M. Monot, M. Vicic, and F. Flament, “Syn- chronized in vivo measurements of skin hydration and trans-epidermal water loss. exploring their mutual influences,”Interna- tional Journal of Cosmetic Science, vol. 41, no. 5, pp. 437–442, 2019

    work page 2019

  2. [2]

    Transepidermal water loss in healthy adults: a systematic review and meta-analysis update,

    M. Akdeniz, S. Gabriel, A. Lichterfeld- Kottner, U. Blume-Peytavi, and J. Kot- tner, “Transepidermal water loss in healthy adults: a systematic review and meta-analysis update,”British Journal of Dermatology, vol. 179, no. 5, pp. 1049–1055, 2018

    work page 2018

  3. [3]

    Voegeli, J

    R. Voegeli, J. Gierschendorf, B. Summers, and A. Rawlings, “Facial skin mapping: from single point bio-instrumental evaluation to continuous visualization of skin hydration, barrier function, skin surface ph, and sebum in different ethnic skin types,”International Springer Nature 2021 LATEX template AI-driven Remote Facial Skin Hydration and TEWL Assessme...

    work page 2021

  4. [4]

    Skin hydra- tion: a review on its molecular mechanisms,

    S. Verdier-S´ evrain and F. Bont´ e, “Skin hydra- tion: a review on its molecular mechanisms,” Journal of cosmetic dermatology, vol. 6, no. 2, pp. 75–82, 2007

    work page 2007

  5. [5]

    Enhanced fluctu- ations in facial pore size, redness, and tewl caused by mask usage are normalized by the application of a moisturizer,

    K. Miyamoto, Y. Munakata, X. Yan, G. Tsuji, and M. Furue, “Enhanced fluctu- ations in facial pore size, redness, and tewl caused by mask usage are normalized by the application of a moisturizer,”Journal of Clinical Medicine, vol. 11, no. 8, p. 2121, 2022

    work page 2022

  6. [6]

    Assessment of skin barrier function using skin images with topological data analysis,

    K. Koseki, H. Kawasaki, T. Atsugi, M. Nakanishi, M. Mizuno, E. Naru, T. Ebi- hara, M. Amagai, and E. Kawakami, “Assessment of skin barrier function using skin images with topological data analysis,” NPJ systems biology and applications, vol. 6, no. 1, p. 40, 2020

    work page 2020

  7. [7]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly,et al., “An image is worth 16x16 words: Transformers for image recognition at scale,”arXiv preprint arXiv:2010.11929, 2020

    arXiv 2010

  8. [8]

    Explicit visual prompting for low-level structure segmentations,

    W. Liu, X. Shen, C.-M. Pun, and X. Cun, “Explicit visual prompting for low-level structure segmentations,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 19434– 19445, 2023

    work page 2023

Show all 58 references
  1. [9]

    Visual prompt tuning,

    M. Jia, L. Tang, B.-C. Chen, C. Cardie, S. Belongie, B. Hariharan, and S.-N. Lim, “Visual prompt tuning,” inEuropean Con- ference on Computer Vision, pp. 709–727, Springer, 2022

    2022

  2. [10]

    Vision transformer adapter for dense predictions,

    Z. Chen, Y. Duan, W. Wang, J. He, T. Lu, J. Dai, and Y. Qiao, “Vision transformer adapter for dense predictions,”arXiv preprint arXiv:2205.08534, 2022

    2022 arXiv

  3. [11]

    Convolutional bypasses are better vision transformer adapters,

    S. Jie and Z.-H. Deng, “Convolutional bypasses are better vision transformer adapters,”arXiv preprint arXiv:2207.07039, 2022

    2022 arXiv

  4. [12]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on com- puter vision and pattern recognition, pp. 770– 778, 2016

    2016

  5. [13]

    Very deep convolutional networks for large- scale image recognition,

    K. Simonyan and A. Zisserman, “Very deep convolutional networks for large- scale image recognition,”arXiv preprint arXiv:1409.1556, 2014

    2014 arXiv

  6. [14]

    Efficientnet: Rethink- ing model scaling for convolutional neural networks,

    M. Tan and Q. Le, “Efficientnet: Rethink- ing model scaling for convolutional neural networks,” inInternational conference on machine learning, pp. 6105–6114, PMLR, 2019

    2019

  7. [15]

    A convnet for the 2020s,

    Z. Liu, H. Mao, C.-Y. Wu, C. Feichten- hofer, T. Darrell, and S. Xie, “A convnet for the 2020s,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 11976–11986, 2022

    2022

  8. [16]

    Swin transformer: Hierarchical vision transformer using shifted windows,

    Z. Liu, Y. Lin, Y. Cao, H. Hu, Y. Wei, Z. Zhang, S. Lin, and B. Guo, “Swin transformer: Hierarchical vision transformer using shifted windows,” inProceedings of the IEEE/CVF international conference on computer vision, pp. 10012–10022, 2021

    2021

  9. [17]

    Pyra- mid vision transformer: A versatile backbone for dense prediction without convolutions,

    W. Wang, E. Xie, X. Li, D.-P. Fan, K. Song, D. Liang, T. Lu, P. Luo, and L. Shao, “Pyra- mid vision transformer: A versatile backbone for dense prediction without convolutions,” in Proceedings of the IEEE/CVF international conference on computer vision, pp. 568–578, 2021

    2021

  10. [18]

    Mul- ticentre comparison of skin hydration in terms of physical-, physiological-and product- dependent parameters by the capacitive method (corneometer cm 825),

    U. Heinrich, U. Koop, M.-C. Leneveu- Duchemin, K. Osterrieder, S. Bielfeldt, C. Chkarnat, J. Degwert, D. H¨ antschel, S. Jaspers, H.-P. Nissen,et al., “Mul- ticentre comparison of skin hydration in terms of physical-, physiological-and product- dependent parameters by the capa...

    2003

  11. [19]

    Devices measuring transepidermal water loss: A sys- tematic review of measurement properties,

    T. Klotz, A. Ibrahim, G. Maddern, Y. Caplash, and M. Wagstaff, “Devices measuring transepidermal water loss: A sys- tematic review of measurement properties,” Skin Research and Technology, vol. 28, no. 4, pp. 497–539, 2022

    2022

  12. [20]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszko- reit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in neural information processing systems, vol. 30, 2017

    2017

  13. [21]

    Bert: Pre-training of deep bidirectional transformers for language understanding,

    J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “Bert: Pre-training of deep bidirectional transformers for language understanding,” inProceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologie...

    2019

  14. [22]

    Exploring the limits of transfer learning with a unified text-to-text transformer,

    C. Raffel, N. Shazeer, A. Roberts, K. Lee, S. Narang, M. Matena, Y. Zhou, W. Li, and P. J. Liu, “Exploring the limits of transfer learning with a unified text-to-text transformer,”Journal of machine learning research, vol. 21, no. 140, pp. 1–67, 2020

    2020

  15. [23]

    Language models are few-shot learners,

    T. Brown, B. Mann, N. Ryder, M. Subbiah, J. D. Kaplan, P. Dhariwal, A. Neelakan- tan, P. Shyam, G. Sastry, A. Askell,et al., “Language models are few-shot learners,” Advances in neural information processing systems, vol. 33, pp. 1877–1901, 2020

    1901

  16. [24]

    Llama: Open and efficient foun- dation language models,

    H. Touvron, T. Lavril, G. Izacard, X. Martinet, M.-A. Lachaux, T. Lacroix, B. Rozi` ere, N. Goyal, E. Hambro, F. Azhar, et al., “Llama: Open and efficient foun- dation language models,”arXiv preprint arXiv:2302.13971, 2023

    2023 arXiv

  17. [25]

    Mvitv2: Improved multiscale vision trans- formers for classification and detection,

    Y. Li, C.-Y. Wu, H. Fan, K. Mangalam, B. Xiong, J. Malik, and C. Feichtenhofer, “Mvitv2: Improved multiscale vision trans- formers for classification and detection,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 4804–4814, 2022

    2022

  18. [26]

    Ima- genet large scale visual recognition chal- lenge,

    O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpa- thy, A. Khosla, M. Bernstein,et al., “Ima- genet large scale visual recognition chal- lenge,”International journal of computer vision, vol. 115, pp. 211–252, 2015

    2015

  19. [27]

    Parameter-efficient transfer learning for nlp,

    N. Houlsby, A. Giurgiu, S. Jastrzebski, B. Morrone, Q. De Laroussilhe, A. Ges- mundo, M. Attariyan, and S. Gelly, “Parameter-efficient transfer learning for nlp,” inInternational conference on machine learning, pp. 2790–2799, PMLR, 2019

    2019

  20. [28]

    Towards a unified view of parameter-efficient transfer learning,

    J. He, C. Zhou, X. Ma, T. Berg-Kirkpatrick, and G. Neubig, “Towards a unified view of parameter-efficient transfer learning,”arXiv preprint arXiv:2110.04366, 2021

    2021 arXiv

  21. [29]

    Lora: Low-rank adaptation of large language models.,

    E. J. Hu, Y. Shen, P. Wallis, Z. Allen-Zhu, Y. Li, S. Wang, L. Wang, W. Chen,et al., “Lora: Low-rank adaptation of large language models.,”ICLR, vol. 1, no. 2, p. 3, 2022

    2022

  22. [30]

    Prefix-tuning: Opti- mizing continuous prompts for generation,

    X. L. Li and P. Liang, “Prefix-tuning: Opti- mizing continuous prompts for generation,” arXiv preprint arXiv:2101.00190, 2021

    2021 arXiv

  23. [31]

    The power of scale for parameter-efficient prompt tuning,

    B. Lester, R. Al-Rfou, and N. Constant, “The power of scale for parameter-efficient prompt tuning,”arXiv preprint arXiv:2104.08691, 2021

    2021 arXiv

  24. [32]

    Gpt understands, too,

    X. Liu, Y. Zheng, Z. Du, M. Ding, Y. Qian, Z. Yang, and J. Tang, “Gpt understands, too,”AI Open, vol. 5, pp. 208–215, 2024

    2024

  25. [33]

    P-tuning v2: Prompt tuning can be comparable to fine-tuning universally across scales and tasks,

    X. Liu, K. Ji, Y. Fu, W. L. Tam, Z. Du, Z. Yang, and J. Tang, “P-tuning v2: Prompt tuning can be comparable to fine-tuning universally across scales and tasks,”arXiv preprint arXiv:2110.07602, 2021

    2021 arXiv

  26. [34]

    Llama-adapter: Efficient fine-tuning of large language mod- els with zero-initialized attention,

    R. Zhang, J. Han, C. Liu, A. Zhou, P. Lu, Y. Qiao, H. Li, and P. Gao, “Llama-adapter: Efficient fine-tuning of large language mod- els with zero-initialized attention,” inThe Twelfth International Conference on Learn- ing Representations, 2024

    2024

  27. [35]

    A. Luo, R. Cai, C. Kong, Y. Ju, X. Kang, J. Huang, and A. C. K. Life, “Forgery-aware Springer Nature 2021 LATEX template AI-driven Remote Facial Skin Hydration and TEWL Assessment from Selfie Images: A Systematic Solution21 adaptive learning with vision transformer for general...

    2021

  28. [36]

    Towards data-centric face anti-spoofing: Improving cross-domain gener- alization via physics-based data synthesis,

    R. Cai, C. Soh, Z. Yu, H. Li, W. Yang, and A. C. Kot, “Towards data-centric face anti-spoofing: Improving cross-domain gener- alization via physics-based data synthesis,” International Journal of Computer Vision, pp. 1–22, 2024

    2024

  29. [37]

    Suppress and rebalance: Towards generalized multi-modal face anti-spoofing,

    X. Lin, S. Wang, R. Cai, Y. Liu, Y. Fu, W. Tang, Z. Yu, and A. Kot, “Suppress and rebalance: Towards generalized multi-modal face anti-spoofing,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 211–221, 2024

    2024

  30. [38]

    Visual prompt flexible-modal face anti- spoofing,

    Z. Yu, R. Cai, Y. Cui, A. Liu, and C. Chen, “Visual prompt flexible-modal face anti- spoofing,”IEEE Transactions on Dependable and Secure Computing, 2024

    2024

  31. [39]

    S-adapter: General- izing vision transformer for face anti-spoofing with statistical tokens,

    R. Cai, Z. Yu, C. Kong, H. Li, C. Chen, Y. Hu, and A. C. Kot, “S-adapter: General- izing vision transformer for face anti-spoofing with statistical tokens,”IEEE Transactions on Information Forensics and Security, 2024

    2024

  32. [40]

    Rehearsal-free domain con- tinual face anti-spoofing: Generalize more and forget less,

    R. Cai, Y. Cui, Z. Li, Z. Yu, H. Li, Y. Hu, and A. Kot, “Rehearsal-free domain con- tinual face anti-spoofing: Generalize more and forget less,” inProceedings of the IEEE/CVF International Conference on Computer Vision, pp. 8037–8048, 2023

    2023

  33. [41]

    Smote for regression,

    L. Torgo, R. P. Ribeiro, B. Pfahringer, and P. Branco, “Smote for regression,” inPor- tuguese conference on artificial intelligence, pp. 378–389, Springer, 2013

    2013

  34. [42]

    Smogn: a pre-processing approach for imbalanced regression,

    P. Branco, L. Torgo, and R. P. Ribeiro, “Smogn: a pre-processing approach for imbalanced regression,” inFirst interna- tional workshop on learning with imbalanced domains: Theory and applications, pp. 36–50, PMLR, 2017

    2017

  35. [43]

    Rebagg: Resampled bagging for imbalanced regression,

    P. Branco, L. Torgo, and R. P. Ribeiro, “Rebagg: Resampled bagging for imbalanced regression,” inSecond International Work- shop on Learning with Imbalanced Domains: Theory and Applications, pp. 67–81, PMLR, 2018

    2018

  36. [44]

    Density-based weighting for imbalanced regression,

    M. Steininger, K. Kobs, P. Davidson, A. Krause, and A. Hotho, “Density-based weighting for imbalanced regression,” Machine Learning, vol. 110, pp. 2187–2211, 2021

    2021

  37. [45]

    Delving into deep imbalanced regression,

    Y. Yang, K. Zha, Y. Chen, H. Wang, and D. Katabi, “Delving into deep imbalanced regression,” inInternational conference on machine learning, pp. 11842–11851, PMLR, 2021

    2021

  38. [46]

    Bal- anced mse for imbalanced visual regression,

    J. Ren, M. Zhang, C. Yu, and Z. Liu, “Bal- anced mse for imbalanced visual regression,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion, pp. 7926–7935, 2022

    2022

  39. [47]

    Ranksim: Ranking similarity regularization for deep imbalanced regression,

    Y. Gong, G. Mori, and F. Tung, “Ranksim: Ranking similarity regularization for deep imbalanced regression,”arXiv preprint arXiv:2205.15236, 2022

    2022 arXiv

  40. [48]

    Variational imbal- anced regression: Fair uncertainty quantifica- tion via probabilistic smoothing,

    Z. Wang and H. Wang, “Variational imbal- anced regression: Fair uncertainty quantifica- tion via probabilistic smoothing,”Advances in Neural Information Processing Systems, vol. 36, pp. 30429–30452, 2023

    2023

  41. [49]

    Dad-3dheads: A large-scale dense, accurate and diverse dataset for 3d head alignment from a single image,

    T. Martyniuk, O. Kupyn, Y. Kurlyak, I. Krashenyi, J. Matas, and V. Sharmanska, “Dad-3dheads: A large-scale dense, accurate and diverse dataset for 3d head alignment from a single image,” inProceedings of the IEEE/CVF Conference on computer vision and pattern recognition, pp. 2...

    2022

  42. [50]

    Pointnet: Deep learning on point sets for 3d classification and segmentation,

    C. R. Qi, H. Su, K. Mo, and L. J. Guibas, “Pointnet: Deep learning on point sets for 3d classification and segmentation,” inPro- ceedings of the IEEE conference on computer vision and pattern recognition, pp. 652–660, 2017

    2017

  43. [51]

    Pillow (python imaging library)

    A. Clark and Contributors, “Pillow (python imaging library).” https://python-pillow.org, Springer Nature 2021 LATEX template 22AI-driven Remote Facial Skin Hydration and TEWL Assessment from Selfie Images: A Systematic Solution

    2021

  44. [52]

    Supervised contrastive learn- ing,

    P. Khosla, P. Teterwak, C. Wang, A. Sarna, Y. Tian, P. Isola, A. Maschinot, C. Liu, and D. Krishnan, “Supervised contrastive learn- ing,”Advances in neural information pro- cessing systems, vol. 33, pp. 18661–18673, 2020

    2020

  45. [53]

    Uti- lization of individual typology angle (ita) and hue angle in the measurement of skin color on images,

    Y. Wu, T. Tanaka, and M. Akimoto, “Uti- lization of individual typology angle (ita) and hue angle in the measurement of skin color on images,”bioimages, vol. 28, pp. 1–8, 2020

    2020

  46. [54]

    Fusionmamba: Dynamic feature enhance- ment for multimodal image fusion with mamba,

    X. Xie, Y. Cui, T. Tan, X. Zheng, and Z. Yu, “Fusionmamba: Dynamic feature enhance- ment for multimodal image fusion with mamba,”Visual Intelligence, vol. 2, no. 1, p. 37, 2024

    2024

  47. [55]

    Automatic facial skin fea- ture detection for everyone,

    Q. Zheng, A. Purwar, H. Zhao, G. L. Lim, L. Li, D. Behera, Q. Wang, M. Tan, R. Cai, J. Werner, D. Sng, M. van Steensel, W. Lin, and A. C. Kot, “Automatic facial skin fea- ture detection for everyone,” inProc. IS&T Int’l. Symp. on Electronic Imaging: Imag- ing and Multimedia An...

    2022

  48. [56]

    Evaluating the efficacy of skincare product: A realistic short-term facial pore simulation,

    L. Li, B. Dissanayake, T. Omotezako, Y. Zhong, Q. Zhang, R. Cai, Q. Zheng, D. Sng, W. Lin, Y. Wang, and A. C. Kot, “Evaluating the efficacy of skincare product: A realistic short-term facial pore simulation,” Electronic Imaging, pp. 276–1–276–6, 2023

    2023

  49. [57]

    Controllable and gradual facial blemishes retouching via physics-based modelling,

    C. Shuai, R. Cai, B. Dissanayake, A. New- man, D. Guan, D. Sng, L. Li, and A. Kot, “Controllable and gradual facial blemishes retouching via physics-based modelling,” in 2024 IEEE International Conference on Mul- timedia and Expo (ICME), pp. 1–6, 2024. Biography Cecelia Sohrec...

    2024

  50. [2024]

    His research interests include computer vision and bio- metric/AI security

    Now, he is a research fellow in the Rapid- Rich Object Search (ROSE) Lab and NTU-PKU Joint Research Institute, leading computer vision projects collaborated with industrial partners. His research interests include computer vision and bio- metric/AI security. E-mail: rizhao001@...

    2021

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.