pith. sign in

arxiv: 2412.10679 · v3 · submitted 2024-12-14 · 💻 cs.CV · eess.IV

U-FaceBP: Uncertainty-aware Bayesian Ensemble Deep Learning for Face Video-based Blood Pressure Estimation

Yusuke Akamatsu , Akinori F. Ebihara , Terumi Umematsu This is my paper

Pith reviewed 2026-05-23 07:17 UTC · model grok-4.3

classification 💻 cs.CV eess.IV
keywords blood pressure estimationremote photoplethysmographyBayesian neural networksuncertainty estimationface videoensemble learningmodality fusion
0
0 comments X

The pith

An uncertainty-aware Bayesian ensemble estimates blood pressure from face videos by combining rPPG signals, derived PPG signals, and face images.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents U-FaceBP to handle uncertainties that limit reliability in camera-based blood pressure measurement. It models both aleatoric and epistemic uncertainty through Bayesian neural networks and runs an ensemble across three input types extracted from the same face video. Experiments on two datasets covering 1197 subjects from multiple racial groups show better accuracy than prior methods. The uncertainty outputs are used to weight the different modalities during fusion and to flag less reliable predictions.

Core claim

U-FaceBP is an uncertainty-aware Bayesian ensemble deep learning method for face video-based blood pressure estimation. It employs multiple Bayesian neural networks to process rPPG signals, PPG signals derived from face videos, and face images directly, thereby capturing aleatoric and epistemic uncertainties. Large-scale experiments on two datasets involving 1197 subjects from diverse racial groups demonstrate that U-FaceBP outperforms state-of-the-art BP estimation methods. The uncertainty estimates are shown to be informative for guiding modality fusion, assessing prediction reliability, and analyzing performance differences across racial groups.

What carries the argument

Bayesian ensemble of multiple BNNs that separately process rPPG signals, derived PPG signals, and raw face images to produce uncertainty-aware blood pressure estimates and enable uncertainty-guided fusion.

If this is right

  • The method achieves higher accuracy than existing blood pressure estimation approaches on datasets with 1197 subjects spanning multiple racial groups.
  • Uncertainty estimates can be used to decide how much weight to give each input modality during fusion.
  • Uncertainty values allow identification of predictions that are likely to be less reliable.
  • Performance differences across racial groups can be examined using the same uncertainty outputs.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same uncertainty-guided fusion could be applied to estimate other physiological signals such as heart rate or respiration rate from video.
  • Consumer devices could incorporate this approach to provide continuous non-contact monitoring with built-in reliability indicators.
  • Training on more balanced racial data might further reduce group-level performance gaps when uncertainty is used to audit outputs.

Load-bearing premise

The uncertainty values produced by the Bayesian ensemble are informative enough to improve modality fusion and to indicate which predictions are less reliable across different racial groups.

What would settle it

On the same two datasets, weighting the ensemble members by their reported uncertainty produces no accuracy gain over equal-weight or fixed fusion, or the reported uncertainty shows no correlation with actual absolute error.

Figures

Figures reproduced from arXiv: 2412.10679 by Akinori F. Ebihara, Terumi Umematsu, Yusuke Akamatsu.

Figure 1
Figure 1. Figure 1: (a) PTT- and PWA-based methods for cuffless BP measurement [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Framework of U-FaceBP: (a) BP estimation from rPPG signals, (b) BP estimation from PPG signals, (c) BP estimation from face images, and (d) [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: SBP and DBP distributions in FaceBP dataset. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Correlation plots of SBP and DBP in U-FaceBP. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Correlation plots for SBP and DBP predictions by U-FaceBP with aleatoric, epistemic, and total uncertainties on FaceBP dataset. First and second [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: MAE and total uncertainty for each racial group. “Other Asian” [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: rPPG signals from cheek and forehead in FaceBP dataset. [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Gender and age distribution of subjects in FaceBP dataset. [PITH_FULL_IMAGE:figures/full_fig_p009_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Number of subjects for each racial group in FaceBP dataset. [PITH_FULL_IMAGE:figures/full_fig_p009_10.png] view at source ↗
Figure 12
Figure 12. Figure 12: Correlation plots with total uncertainty for rPPG, PPG, and face image modalities. [PITH_FULL_IMAGE:figures/full_fig_p010_12.png] view at source ↗
read the original abstract

Blood pressure (BP) measurement is crucial for daily health assessment. Remote photoplethysmography (rPPG), which extracts pulse waves from face videos captured by a camera, has the potential to enable convenient BP measurement without specialized medical devices. However, there are various uncertainties in BP estimation using rPPG, leading to limited estimation performance and reliability. In this paper, we propose U-FaceBP, an uncertainty-aware Bayesian ensemble deep learning method for face video-based BP estimation. U-FaceBP models aleatoric and epistemic uncertainties in face video-based BP estimation with a Bayesian neural network (BNN). Additionally, we design U-FaceBP as an ensemble method, estimating BP from rPPG signals, PPG signals derived from face videos, and face images using multiple BNNs. Large-scale experiments on two datasets involving 1197 subjects from diverse racial groups demonstrate that U-FaceBP outperforms state-of-the-art BP estimation methods. Furthermore, we show that the uncertainty estimates provided by U-FaceBP are informative and useful for guiding modality fusion, assessing prediction reliability, and analyzing performance across racial groups.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces U-FaceBP, an uncertainty-aware Bayesian ensemble deep learning method for blood pressure estimation from face videos. It employs multiple Bayesian neural networks to process rPPG signals, derived PPG signals, and face images while modeling both aleatoric and epistemic uncertainties. The central empirical claim is that large-scale experiments on two datasets involving 1197 subjects from diverse racial groups demonstrate outperformance over state-of-the-art BP estimation methods, with the uncertainty estimates shown to be informative for guiding modality fusion, assessing prediction reliability, and analyzing performance across racial groups.

Significance. If the empirical results prove robust, the work contributes to remote photoplethysmography-based BP estimation by incorporating explicit uncertainty modeling within an ensemble framework. The scale of the evaluation (1197 subjects across racial groups) is a positive aspect for an applied ML paper in this area.

major comments (2)
  1. [Results] Results section (performance tables): The reported outperformance metrics lack error bars, standard deviations across runs or folds, or statistical significance tests. This is load-bearing for the central claim of superiority on the 1197-subject datasets, as it prevents assessment of whether observed gains exceed variability.
  2. [§4] §4 (experimental setup): Dataset splits, subject exclusion criteria, and cross-validation details are insufficiently specified. These are necessary to substantiate the generalizability claim across diverse racial groups and to rule out data leakage or selection bias in the large-scale experiments.
minor comments (2)
  1. [§3] The description of how uncertainty estimates are aggregated across the three modalities (rPPG, derived PPG, face images) could be clarified with an explicit equation or pseudocode.
  2. [Figures] Figure captions for uncertainty visualizations should include quantitative metrics (e.g., correlation between uncertainty and absolute error) to support the claim of informativeness.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and will revise the manuscript to improve clarity and robustness of the reported results.

read point-by-point responses

  1. Referee: [Results] Results section (performance tables): The reported outperformance metrics lack error bars, standard deviations across runs or folds, or statistical significance tests. This is load-bearing for the central claim of superiority on the 1197-subject datasets, as it prevents assessment of whether observed gains exceed variability.

    Authors: We agree that the absence of variability measures and significance tests limits the strength of the superiority claims. In the revised version, we will add error bars (standard deviations across multiple runs or cross-validation folds) to all performance tables and include statistical significance tests (e.g., paired t-tests with p-values) comparing U-FaceBP against baselines. revision: yes

  2. Referee: [§4] §4 (experimental setup): Dataset splits, subject exclusion criteria, and cross-validation details are insufficiently specified. These are necessary to substantiate the generalizability claim across diverse racial groups and to rule out data leakage or selection bias in the large-scale experiments.

    Authors: We acknowledge the need for greater transparency. The revised Section 4 will explicitly detail the subject-independent train/validation/test splits, subject exclusion criteria (e.g., based on video quality or missing metadata), and the cross-validation protocol, confirming no data leakage and supporting the generalizability claims across racial groups. revision: yes

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The paper is an empirical ML study whose central claim is outperformance of a Bayesian ensemble method on two external datasets (1197 subjects). No mathematical derivation chain, fitted-parameter-as-prediction, self-definitional loop, or load-bearing self-citation is present; the reported results are obtained by training and evaluating networks on held-out data and comparing to independent SOTA baselines. The uncertainty modeling is presented as an auxiliary output rather than the source of the performance numbers themselves. The derivation is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The method rests on standard deep-learning assumptions plus several fitted components typical of BNN ensembles; no independent evidence is supplied for the utility of the uncertainty estimates beyond the reported experiments.

free parameters (1)
  • BNN hyperparameters and ensemble weights
    Network architecture, prior distributions, and fusion weights are chosen or optimized on the training data.
axioms (1)
  • domain assumption Bayesian neural networks can separately capture aleatoric and epistemic uncertainty in rPPG signals
    Invoked when the paper states that BNNs model both uncertainty types for BP estimation.

pith-pipeline@v0.9.0 · 5740 in / 1313 out tokens · 28122 ms · 2026-05-23T07:17:23.617940+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

76 extracted references · 76 canonical work pages · 3 internal anchors

  1. [1]

    World health statistics 2022: monitoring health for the SDGs, sustainable development goals,

    World Health Organization, “World health statistics 2022: monitoring health for the SDGs, sustainable development goals,” 2022

    work page 2022

  2. [2]

    Factors associated with the level of knowledge about hypertension in primary care patients,

    ´AR Lugo-Mata, AS Urich-Landeta, AL Andrades-P ´erez, MJ Le ´on- Dugarte, et al., “Factors associated with the level of knowledge about hypertension in primary care patients,”Medicina Universitaria, vol. 19, no. 77, pp. 184–188, 2017

    work page 2017

  3. [3]

    Noninvasive and cuffless measurements of blood pressure for telemedicine,

    KW Chan, K Hung, and YT Zhang, “Noninvasive and cuffless measurements of blood pressure for telemedicine,” inProc. Annual Int. Conf. IEEE Engineering in Medicine and Biology Society (EMBC), 2001, vol. 4, pp. 3592–3593

    work page 2001

  4. [4]

    Continuous blood pressure measurement by using the pulse transit time: comparison to a cuff-based method,

    H. Gesche, D. Grosskurth, G. K ¨uchler, and A. Patzak, “Continuous blood pressure measurement by using the pulse transit time: comparison to a cuff-based method,”European Journal of Applied Physiology, vol. 112, no. 1, pp. 309–315, 2012

    work page 2012

  5. [5]

    Cuffless blood pressure estimation algorithms for continuous health- care monitoring,

    M. Kachuee, M. M. Kiani, H. Mohammadzade, and M. Shabany, “Cuffless blood pressure estimation algorithms for continuous health- care monitoring,”IEEE Transactions on Biomedical Engineering, vol. 64, no. 4, pp. 859–869, 2016

    work page 2016

  6. [6]

    Estimation and validation of arterial blood pressure using photoplethysmogram morphology features in conjunction with pulse arrival time in large open databases,

    S. Yang, J. Sohn, S. Lee, J. Lee, et al., “Estimation and validation of arterial blood pressure using photoplethysmogram morphology features in conjunction with pulse arrival time in large open databases,”IEEE Journal of Biomedical and Health Informatics, vol. 25, no. 4, pp. 1018– 1030, 2020

    work page 2020

  7. [7]

    Con- ventional pulse transit times as markers of blood pressure changes in humans,

    R. C. Block, M. Yavarimanesh, K. Natarajan, A. Carek, et al., “Con- ventional pulse transit times as markers of blood pressure changes in humans,”Scientific Reports, vol. 10, no. 1, pp. 16373, 2020

    work page 2020

  8. [8]

    Continuous and noninvasive estimation of arterial blood pressure using a photoplethysmographic approach,

    XF Teng and YT Zhang, “Continuous and noninvasive estimation of arterial blood pressure using a photoplethysmographic approach,” in Proc. Annual Int. Conf. IEEE Engineering in Medicine and Biology Society (EMBC), 2003, vol. 4, pp. 3153–3156

    work page 2003

  9. [9]

    A neural network-based method for continuous blood pressure estimation from a PPG signal,

    Y . Kurylyak, F. Lamonaca, and D. Grimaldi, “A neural network-based method for continuous blood pressure estimation from a PPG signal,” inProc. Int. Instrumentation and Measurement Technology Conference (I2MTC), 2013, pp. 280–283

    work page 2013

  10. [10]

    Optical blood pressure estimation with photo- plethysmography and FFT-based neural networks,

    X. Xing and M. Sun, “Optical blood pressure estimation with photo- plethysmography and FFT-based neural networks,”Biomedical Optics Express, vol. 7, no. 8, pp. 3007–3020, 2016

    work page 2016

  11. [11]

    Blood pressure estimation from photoplethysmogram using a spectro-temporal deep neural net- work,

    G. Slapni ˇcar, N. Mlakar, and M. Lu ˇstrek, “Blood pressure estimation from photoplethysmogram using a spectro-temporal deep neural net- work,”Sensors, vol. 19, no. 15, pp. 3420, 2019

    work page 2019

  12. [12]

    Personalized blood pressure estimation using photoplethysmography: A transfer learning approach,

    J. Leitner, P. Chiang, and S. Dey, “Personalized blood pressure estimation using photoplethysmography: A transfer learning approach,” IEEE Journal of Biomedical and Health Informatics, vol. 26, no. 1, pp. 218–228, 2021

    work page 2021

  13. [13]

    Pulse transit time as an indicator of arterial blood pressure,

    LA Geddes, MH V oelz, CF Babbs, JD Bourland, et al., “Pulse transit time as an indicator of arterial blood pressure,”Psychophysiology, vol. 18, no. 1, pp. 71–74, 1981

    work page 1981

  14. [14]

    Blood pressure measurement using only a smartphone,

    L. Frey, C. Menon, and M. Elgendi, “Blood pressure measurement using only a smartphone,”NPJ Digital Medicine, vol. 5, no. 1, pp. 86, 2022

    work page 2022

  15. [15]

    Blood pressure measurement: From cuff-based to contactless monitoring,

    P. Man, K. Cheung, N. Sangsiri, W. J. Shek, et al., “Blood pressure measurement: From cuff-based to contactless monitoring,”Healthcare, vol. 10, no. 10, pp. 2113, 2022

    work page 2022

  16. [16]

    Smartphone- based blood pressure measurement using transdermal optical imaging technology,

    H. Luo, D. Yang, A. Barszczyk, N. Vempala, et al., “Smartphone- based blood pressure measurement using transdermal optical imaging technology,”Circulation: Cardiovascular Imaging, vol. 12, no. 8, pp. e008857, 2019

    work page 2019

  17. [17]

    The noninvasive blood pressure measurement based on facial images processing,

    Y . Zhou, H. Ni, Q. Zhang, and Q. Wu, “The noninvasive blood pressure measurement based on facial images processing,”IEEE Sensors Journal, vol. 19, no. 22, pp. 10624–10634, 2019

    work page 2019

  18. [18]

    A blood pressure prediction method based on imaging photoplethysmography in combination with machine learning,

    M. Rong and K. Li, “A blood pressure prediction method based on imaging photoplethysmography in combination with machine learning,” Biomedical Signal Processing and Control, vol. 64, pp. 102328, 2021

    work page 2021

  19. [19]

    Assessment of deep learning based blood pressure prediction from PPG and rPPG signals,

    F. Schrumpf, P. Frenzel, C. Aust, G. Osterhoff, et al., “Assessment of deep learning based blood pressure prediction from PPG and rPPG signals,” inProc. IEEE/CVF Conf. Computer Vision and Pattern Recognition Workshop (CVPRW), 2021, pp. 3820–3830

    work page 2021

  20. [20]

    A facial-image-based blood pressure measurement system without calibration,

    B. Wu, B. Wu, B. Tsai, and C. Hsu, “A facial-image-based blood pressure measurement system without calibration,”IEEE Transactions on Instrumentation and Measurement, vol. 71, pp. 1–13, 2022

    work page 2022

  21. [21]

    Hybrid D1DCnet using forehead iPPG for continuous and noncontact blood pressure measurement,

    Y . Li, M. Wei, Q. Chen, X. Zhu, et al., “Hybrid D1DCnet using forehead iPPG for continuous and noncontact blood pressure measurement,”IEEE Sensors Journal, vol. 23, no. 3, pp. 2727–2736, 2022

    work page 2022

  22. [22]

    Camera-based blood pressure estimation via windkessel model and waveform features,

    B. Wu, B. Wu, and C. Hsu, “Camera-based blood pressure estimation via windkessel model and waveform features,”IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1–13, 2022

    work page 2022

  23. [23]

    VidBP: Detecting blood pressure from facial videos with personalized calibration,

    X. Liu, Z. Sun, X. Li, R. Song, et al., “VidBP: Detecting blood pressure from facial videos with personalized calibration,” inProc. Annual Int. Conf. IEEE Engineering in Medicine and Biology Society (EMBC), 2023, pp. 1–5

    work page 2023

  24. [24]

    Assessing heart rate and blood pressure estimation from image photo- plethysmography using a digital blood pressure meter,

    S. Trirongjitmoah, A. Promking, K. Kaewdang, N. Phansiri, et al., “Assessing heart rate and blood pressure estimation from image photo- plethysmography using a digital blood pressure meter,”Heliyon, 2024

    work page 2024

  25. [25]

    Remote blood pressure estimation from facial videos using transfer learning: Leveraging PPG to rPPG conversion,

    C. Cheng, J. W. Chin, K. L. Wong, T. T. Chan, et al., “Remote blood pressure estimation from facial videos using transfer learning: Leveraging PPG to rPPG conversion,” inIEEE/CVF Winter Conf. Applications of Computer Vision (WACV), 2025, pp. 4225–4236

    work page 2025

  26. [26]

    Contactless blood pressure measurement via remote photoplethysmography with synthetic data generation using generative adversarial networks,

    B. Wu, L. Chiu, Y . Wu, C. Lai, et al., “Contactless blood pressure measurement via remote photoplethysmography with synthetic data generation using generative adversarial networks,”IEEE Journal of Biomedical and Health Informatics, vol. 28, no. 2, pp. 621–632, 2023

    work page 2023

  27. [27]

    Robust blood pressure measurement from facial videos in diverse environments,

    J. Park and K. Hong, “Robust blood pressure measurement from facial videos in diverse environments,”Heliyon, vol. 10, no. 4, 2024

    work page 2024

  28. [28]

    Mechanical factors in arterial aging: a clinical perspective,

    M. F. O’rourke and J. Hashimoto, “Mechanical factors in arterial aging: a clinical perspective,”Journal of the American College of Cardiology, vol. 50, no. 1, pp. 1–13, 2007

    work page 2007

  29. [29]

    Change in body mass index and its impact on blood pressure: a prospective population study,

    WB. Drøyvold, K. Midthjell, TIL. Nilsen, and J. Holmen, “Change in body mass index and its impact on blood pressure: a prospective population study,”International Journal of Obesity, vol. 29, no. 6, pp. 650–655, 2005

    work page 2005

  30. [30]

    Examining the chal- lenges of blood pressure estimation via photoplethysmogram,

    S. Mehta, N. Kwatra, M. Jain, and D. McDuff, “Examining the chal- lenges of blood pressure estimation via photoplethysmogram,”Scientific Reports, vol. 14, no. 1, pp. 18318, 2024

    work page 2024

  31. [31]

    Estimation of health conditions from facial videos based on deep learning: Addressing limited quantity and quality of human data,

    Y . Akamatsu, “Estimation of health conditions from facial videos based on deep learning: Addressing limited quantity and quality of human data,”Ph.D. Thesis, University of Tsukuba, Japan, 2025

    work page 2025

  32. [32]

    Algorithmic principles of remote PPG,

    W. Wang, A. C. Den Brinker, S. Stuijk, and G. De Haan, “Algorithmic principles of remote PPG,”IEEE Transactions on Biomedical Engineer- ing, vol. 64, no. 7, pp. 1479–1491, 2016

    work page 2016

  33. [33]

    Facial shape analysis identifies valid cues to aspects of physiological health in caucasian, asian, and african populations,

    I. D. Stephen, V . Hiew, V . Coetzee, B. P. Tiddeman, et al., “Facial shape analysis identifies valid cues to aspects of physiological health in caucasian, asian, and african populations,”Frontiers in Psychology, vol. 8, pp. 288503, 2017

    work page 2017

  34. [34]

    A novel method in predicting hypertension using facial images,

    L. Ang, M. H. Yim, J. Do, and S. Lee, “A novel method in predicting hypertension using facial images,”Applied Sciences, vol. 11, no. 5, pp. 2414, 2021

    work page 2021

  35. [35]

    Feasibility of using deep learning to detect coronary artery disease based on facial photo,

    S. Lin, Z. Li, B. Fu, S. Chen, et al., “Feasibility of using deep learning to detect coronary artery disease based on facial photo,”European Heart Journal, vol. 41, no. 46, pp. 4400–4411, 2020

    work page 2020

  36. [36]

    VideoCAD: an uncertainty- driven neural network for coronary artery disease screening from facial videos,

    X. Liu, X. Yang, R. Song, J. Zhang, et al., “VideoCAD: an uncertainty- driven neural network for coronary artery disease screening from facial videos,”IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1–12, 2022

    work page 2022

  37. [37]

    Non-contact, automated cardiac pulse measurements using video imaging and blind source separation.,

    M. Poh, D. J. McDuff, and R. W. Picard, “Non-contact, automated cardiac pulse measurements using video imaging and blind source separation.,”Optics Express, vol. 18, no. 10, pp. 10762–10774, 2010

    work page 2010

  38. [38]

    Robust pulse rate from chrominance-based rPPG,

    G. De Haan and V . Jeanne, “Robust pulse rate from chrominance-based rPPG,”IEEE Transactions on Biomedical Engineering, vol. 60, no. 10, pp. 2878–2886, 2013

    work page 2013

  39. [39]

    Noncontact blood pressure estimation using BP-related cardiovascular knowledge: An uncalibrated method based on consumer-level camera,

    X. Han, X. Yang, S. Fang, R. Song, et al., “Noncontact blood pressure estimation using BP-related cardiovascular knowledge: An uncalibrated method based on consumer-level camera,”IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1–13, 2023. 12

    work page 2023

  40. [40]

    A survey of uncertainty in deep neural networks,

    J. Gawlikowski, C. R. N. Tassi, M. Ali, J. Lee, et al., “A survey of uncertainty in deep neural networks,”Artificial Intelligence Review, vol. 56, no. Suppl 1, pp. 1513–1589, 2023

    work page 2023

  41. [41]

    Exploring uncertainty measures in deep networks for multiple sclerosis lesion detection and segmentation,

    T. Nair, D. Precup, D. L. Arnold, and T. Arbel, “Exploring uncertainty measures in deep networks for multiple sclerosis lesion detection and segmentation,”Medical Image Analysis, vol. 59, pp. 101557, 2020

    work page 2020

  42. [42]

    Towards safe autonomous driving: Capture uncertainty in the deep neural network for lidar 3D vehicle detection,

    D. Feng, L. Rosenbaum, and K. Dietmayer, “Towards safe autonomous driving: Capture uncertainty in the deep neural network for lidar 3D vehicle detection,” inProc. Int. Conf. Intelligent Transportation Systems (ITSC), 2018, pp. 3266–3273

    work page 2018

  43. [43]

    Weight uncertainty in neural network,

    C. Blundell, J. Cornebise, K. Kavukcuoglu, and D. Wierstra, “Weight uncertainty in neural network,” inProc. Int. Conf. Machine Learning (ICML), 2015, pp. 1613–1622

    work page 2015

  44. [44]

    Bayesian Convolutional Neural Networks with Bernoulli Approximate Variational Inference

    Y . Gal and Z. Ghahramani, “Bayesian convolutional neural networks with Bernoulli approximate variational inference,”arXiv preprint arXiv:1506.02158, 2015

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  45. [45]

    What uncertainties do we need in Bayesian deep learning for computer vision?,

    A. Kendall and Y . Gal, “What uncertainties do we need in Bayesian deep learning for computer vision?,”Proc. Advances in Neural Information Processing Systems (NeurIPS), vol. 30, 2017

    work page 2017

  46. [46]

    Bayesian Hypernetworks

    D. Krueger, C. Huang, R. Islam, R. Turner, et al., “Bayesian hypernet- works,”arXiv preprint arXiv:1710.04759, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  47. [47]

    Dropconnect is effective in modeling uncertainty of Bayesian deep networks,

    A. Mobiny, P. Yuan, S. K. Moulik, N. Garg, et al., “Dropconnect is effective in modeling uncertainty of Bayesian deep networks,”Scientific Reports, vol. 11, no. 1, pp. 5458, 2021

    work page 2021

  48. [48]

    Keeping the neural networks simple by minimizing the description length of the weights,

    G. E. Hinton and D. Van Camp, “Keeping the neural networks simple by minimizing the description length of the weights,” inProc. of the Sixth Annual Conf. Computational Learning Theory, 1993, pp. 5–13

    work page 1993

  49. [49]

    Dropout as a Bayesian approximation: Representing model uncertainty in deep learning,

    Y . Gal and Z. Ghahramani, “Dropout as a Bayesian approximation: Representing model uncertainty in deep learning,” inProc. Int. Conf. Machine Learning (ICML), 2016, pp. 1050–1059

    work page 2016

  50. [50]

    Bayesian training of backpropagation networks by the hybrid monte carlo method,

    R. M. Neal, “Bayesian training of backpropagation networks by the hybrid monte carlo method,” Tech. Rep., Technical Report CRG-TR- 92-1, Dept. of Computer Science, University of Toronto, 1992

    work page 1992

  51. [51]

    Transforming neural-net output levels to probability distributions,

    J. Denker and Y . LeCun, “Transforming neural-net output levels to probability distributions,”Proc. Advances in Neural Information Processing Systems (NeurIPS), vol. 3, 1990

    work page 1990

  52. [52]

    Laplace redux-effortless Bayesian deep learning,

    E. Daxberger, A. Kristiadi, A. Immer, et al., “Laplace redux-effortless Bayesian deep learning,”Proc. Advances in Neural Information Pro- cessing Systems (NeurIPS), vol. 34, pp. 20089–20103, 2021

    work page 2021

  53. [53]

    Simple and scalable predictive uncertainty estimation using deep ensembles,

    B. Lakshminarayanan, A. Pritzel, and C. Blundell, “Simple and scalable predictive uncertainty estimation using deep ensembles,”Proc. Advances in Neural Information Processing Systems (NeurIPS), vol. 30, 2017

    work page 2017

  54. [54]

    Deep sub-ensembles for fast uncertainty estima- tion in image classification,

    M. Valdenegro-Toro, “Deep sub-ensembles for fast uncertainty estima- tion in image classification,”arXiv preprint arXiv:1910.08168, 2019

    work page arXiv 1910

  55. [55]

    BatchEnsemble: an alternative approach to efficient ensemble and lifelong learning,

    Yeming Wen, Dustin Tran, and Jimmy Ba, “BatchEnsemble: an alternative approach to efficient ensemble and lifelong learning,” in Proc. Int. Conf. Learning Representations (ICLR), 2019

    work page 2019

  56. [56]

    The future of biometrics technology: from face recognition to related appli- cations,

    H. Imaoka, H. Hashimoto, K. Takahashi, A. F. Ebihara, et al., “The future of biometrics technology: from face recognition to related appli- cations,”APSIPA Transactions on Signal and Information Processing, vol. 10, 2021

    work page 2021

  57. [57]

    DeepPhys: Video-based physiological measurement using convolutional attention networks,

    W. Chen and D. McDuff, “DeepPhys: Video-based physiological measurement using convolutional attention networks,” inProc. European Conference on Computer Vision (ECCV), 2018, pp. 349–365

    work page 2018

  58. [58]

    Multi-task temporal shift attention networks for on-device contactless vitals measurement,

    X. Liu, J. Fromm, S. Patel, and D. McDuff, “Multi-task temporal shift attention networks for on-device contactless vitals measurement,”Proc. Advances in Neural Information Processing Systems (NeurIPS), vol. 33, pp. 19400–19411, 2020

    work page 2020

  59. [59]

    PhysFormer: Facial video- based physiological measurement with temporal difference transformer,

    Z. Yu, Y . Shen, J. Shi, H. Zhao, et al., “PhysFormer: Facial video- based physiological measurement with temporal difference transformer,” inProc. IEEE/CVF Conf. Computer Vision and Pattern Recognition (CVPR), 2022, pp. 4186–4196

    work page 2022

  60. [60]

    CalibrationPhys: Self- supervised video-based heart and respiratory rate measurements by calibrating between multiple cameras,

    Y . Akamatsu, T. Umematsu, and H. Imaoka, “CalibrationPhys: Self- supervised video-based heart and respiratory rate measurements by calibrating between multiple cameras,”IEEE Journal of Biomedical and Health Informatics, vol. 28, pp. 1460–1471, 2024

    work page 2024

  61. [61]

    Cuffless blood pressure estimation based on photoplethysmography signal and its second derivative,

    M. Liu, L. Po, and H. Fu, “Cuffless blood pressure estimation based on photoplethysmography signal and its second derivative,”International Journal of Computer Theory and Engineering, vol. 9, no. 3, pp. 202, 2017

    work page 2017

  62. [62]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inProc. IEEE/CVF Conf. Computer Vision and Pattern Recognition (CVPR), 2016, pp. 770–778

    work page 2016

  63. [63]

    Pre-training strategies and datasets for facial representation learning,

    A. Bulat, S. Cheng, J. Yang, A. Garbett, et al., “Pre-training strategies and datasets for facial representation learning,” inProc. European Conf. Computer Vision (ECCV), 2022, pp. 107–125

    work page 2022

  64. [64]

    Unsupervised learning of visual features by contrasting cluster assignments,

    M. Caron, I. Misra, J. Mairal, P. Goyal, et al., “Unsupervised learning of visual features by contrasting cluster assignments,”Proc. Advances in Neural Information Processing Systems (NeurIPS), vol. 33, pp. 9912– 9924, 2020

    work page 2020

  65. [65]

    Full-body cardiovascular sensing with remote photoplethysmography,

    L. Niu, J. Speth, N. Vance, B. Sporrer, et al., “Full-body cardiovascular sensing with remote photoplethysmography,” inProc. IEEE/CVF Conf. Computer Vision and Pattern Recognition Workshop (CVPRW), 2023, pp. 5994–6004

    work page 2023

  66. [66]

    MSPM: A multisite physiological monitoring dataset for remote pulse, respiration, and blood pressure estimation,

    J. Speth, N. R. Vance, B. Sporrer, L. Niu, et al., “MSPM: A multisite physiological monitoring dataset for remote pulse, respiration, and blood pressure estimation,”IEEE Transactions on Instrumentation and Measurement, vol. 73, pp. 1–14, 2024

    work page 2024

  67. [67]

    Adam: A Method for Stochastic Optimization

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  68. [68]

    PyTorch: An imperative style, high-performance deep learning library,

    A. Paszke, S. Gross, F. Massa, A. Lerer, et al., “PyTorch: An imperative style, high-performance deep learning library,”Proc. Advances in Neural Information Processing Systems (NeurIPS), vol. 32, 2019

    work page 2019

  69. [69]

    Ensemble deep learning for blood pressure estimation using facial videos,

    W. Liu, B. Wu, M. Zhou, X. Zheng, et al., “Ensemble deep learning for blood pressure estimation using facial videos,” inProc. The 3rd Vision-based Remote Physiological Signal Sensing (RePSS) Challenge & Workshop, 2024, pp. 1–10

    work page 2024

  70. [70]

    The 3rd vision-based remote physiological signal sensing (repss) challenge & workshop,

    Z. Sun, X. Li, H. Han, J. Tang, et al., “The 3rd vision-based remote physiological signal sensing (repss) challenge & workshop,” inCEUR Workshop Proceedings, 2024

    work page 2024

  71. [71]

    The British hypertension society protocol for the evaluation of blood pressure measuring devices,

    E. O’Brien, J. Petrie, W. Littler, M. De Swiet, et al., “The British hypertension society protocol for the evaluation of blood pressure measuring devices,”Journal of Hypertension, vol. 11, no. Suppl 2, pp. S43–S62, 1993

    work page 1993

  72. [72]

    A benchmark for machine- learning based non-invasive blood pressure estimation using photo- plethysmogram,

    S. Gonz ´alez, W. Hsieh, and T. P. Chen, “A benchmark for machine- learning based non-invasive blood pressure estimation using photo- plethysmogram,”Scientific Data, vol. 10, no. 1, pp. 149, 2023

    work page 2023

  73. [73]

    Exploring fusion techniques in multimodal AI-based recruitment: Insights from FairCVdb,

    S. Swati, A. Roy, and E. Ntoutsi, “Exploring fusion techniques in multimodal AI-based recruitment: Insights from FairCVdb,”arXiv preprint arXiv:2407.16892, 2024

    work page arXiv 2024

  74. [74]

    ImageNet: A large-scale hierarchical image database,

    J. Deng, W. Dong, R. Socher, L. Li, et al., “ImageNet: A large-scale hierarchical image database,” inProc. IEEE/CVF Conf. Computer Vision and Pattern Recognition (CVPR), 2009, pp. 248–255

    work page 2009

  75. [75]

    A meta-analysis of the impact of skin tone and gender on non-contact photoplethysmog- raphy measurements,

    E. M. Nowara, D. McDuff, and A. Veeraraghavan, “A meta-analysis of the impact of skin tone and gender on non-contact photoplethysmog- raphy measurements,” inProc. IEEE/CVF Conf. Computer Vision and Pattern Recognition Workshop (CVPRW), 2020, pp. 284–285

    work page 2020

  76. [76]

    FairFace: Face attribute dataset for balanced race, gender, and age for bias measurement and mitigation,

    K. Karkkainen and J. Joo, “FairFace: Face attribute dataset for balanced race, gender, and age for bias measurement and mitigation,” inProc. IEEE/CVF Winter Conf. Applications of Computer Vision (WACV), 2021, pp. 1548–1558

    work page 2021