pith. sign in

arxiv: 2401.04560 · v4 · submitted 2024-01-09 · 💻 cs.CV

Phase-shifted remote photoplethysmography for estimating heart rate and blood pressure from facial video

Gyutae Hwang , Sang Jun Lee This is my paper

Pith reviewed 2026-05-24 04:37 UTC · model grok-4.3

classification 💻 cs.CV
keywords remote photoplethysmographyheart rate estimationblood pressure estimationfacial videodeep learningphase discrepancynon-contact monitoringrPPG
0
0 comments X

The pith

Phase discrepancy between acral and facial rPPG signals extracted from video enables estimation of both heart rate and blood pressure.

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

The paper presents a two-stage deep learning framework that first extracts remote photoplethysmography signals from facial and acral video regions and then uses their phase difference to predict blood pressure values. This approach aims to provide contact-free monitoring of heart rate and blood pressure, which could allow continuous tracking without cuffs or electrodes. The method incorporates data augmentation via frame interpolation and a bounded output layer to keep predictions within physiological ranges. On standard datasets it reports lower errors for heart rate than prior video methods.

Core claim

The paper establishes a 2-stage deep learning framework where DRP-Net infers phase-shifted rPPG signals from acral and facial regions to estimate heart rate, and BBP-Net analyzes the phase discrepancy along with temporal features to estimate SBP and DBP within bounded ranges using a scaled sigmoid. Data augmentation via frame interpolation is used to improve heart rate accuracy. On the MMSE-HR dataset, this yields a heart rate MAE of 1.78 BPM, a 34.31% reduction over the recent method, with SBP and DBP MAEs of 10.19 mmHg and 7.09 mmHg; on V4V the corresponding errors are 3.83 BPM, 13.64 mmHg, and 9.4 mmHg.

What carries the argument

The phase discrepancy between acral and facial rPPG signals, which BBP-Net analyzes to regress systolic and diastolic blood pressure beyond heart rate correlation.

If this is right

  • Heart rate can be estimated from facial video with reduced mean absolute error using dual-region rPPG extraction.
  • Systolic and diastolic blood pressure can be inferred from the same video signals without contact sensors.
  • A scaled sigmoid activation constrains blood pressure outputs to predefined physiological ranges.
  • Frame interpolation augmentation improves rPPG signal quality for more accurate heart rate tracking.
  • The framework shows consistent results on both the MMSE-HR and V4V datasets.

Where Pith is reading between the lines

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

  • The method could support continuous monitoring in telehealth or driver-assist scenarios where contact sensors are impractical.
  • Phase timing differences might be further analyzed to derive additional cardiovascular metrics such as pulse transit time.
  • Performance on larger and more diverse populations would be required to establish broader reliability.
  • Smartphone-camera deployment could increase accessibility of frequent checks but would need robustness to variable lighting and motion.

Load-bearing premise

The phase discrepancy between acral and facial rPPG waveforms supplies independent information sufficient to regress continuous SBP and DBP values rather than merely correlating with heart rate.

What would settle it

Train BBP-Net without the phase discrepancy input between the two rPPG signals and measure whether blood pressure estimation error rises to match or exceed a heart-rate-only baseline.

Figures

Figures reproduced from arXiv: 2401.04560 by Gyutae Hwang, Sang Jun Lee.

Figure 1
Figure 1. Figure 1: Overview of the training pipeline of the proposed method. Red and blue texts indicate ground truth and predicted physiological information, [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Data augmentation process of bradycardia and tachycardia samples. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Architecture of DRP-Net. D. Dual remote photoplethysmography network (DRP-Net) In this paper, we propose DRP-Net to estimate phase-shifted rPPG signals from image sequences, and [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Architecture of BBP-Net. In (2), BPd is the estimated blood pressure, and BP max and BP min are heuristic parameters that represent upper and lower bounds of blood pressure. The symbol BP can be either SBP and DBP, and the upper and lower bounds of SBP [ are set to 155 mmHg and 85 mmHg. On the other hand, the upper and lower bounds of DBP \ are set to 95 mmHg and 45 mmHg. In (2), σ, z and τ denote the sigm… view at source ↗
Figure 5
Figure 5. Figure 5: Visualization of rPPG signals (left) and their PSD (right). The upper [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Bland-Altman plot of predicted and reference blood pressure on the [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Predicted heart rate computed from facial rPPG signals and their reference heart rate. In subfigures (a)–(f), Pearson correlation between the predicted [PITH_FULL_IMAGE:figures/full_fig_p010_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Cross skin tone testing. The horizontal axis represents the Fitzpatrick [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗
Figure 10
Figure 10. Figure 10: Visualization of spatial attention αs. The highlighted areas illustrate regions with higher attention scores. [2] S. Aggarwal and K. Pandey, “Early identification of pcos with commonly known diseases: obesity, diabetes, high blood pressure and heart disease using machine learning techniques,” Expert Systems with Applications, vol. 217, p. 119532, 2023. [3] O. Faust, W. Hong, H. W. Loh, S. Xu, R.-S. Tan, S… view at source ↗
read the original abstract

Human health can be critically affected by cardiovascular diseases, such as hypertension, arrhythmias, and stroke. Heart rate and blood pressure are important biometric information for the monitoring of cardiovascular system and early diagnosis of cardiovascular diseases. Existing methods for estimating the heart rate are based on electrocardiography and photoplethyomography, which require contacting the sensor to the skin surface. Moreover, catheter and cuff-based methods for measuring blood pressure cause inconvenience and have limited applicability. Therefore, in this thesis, we propose a vision-based method for estimating the heart rate and blood pressure. This thesis proposes a 2-stage deep learning framework consisting of a dual remote photoplethysmography network (DRP-Net) and bounded blood pressure network (BBP-Net). In the first stage, DRP-Net infers remote photoplethysmography (rPPG) signals for the acral and facial regions, and these phase-shifted rPPG signals are utilized to estimate the heart rate. In the second stage, BBP-Net integrates temporal features and analyzes phase discrepancy between the acral and facial rPPG signals to estimate SBP and DBP values. To improve the accuracy of estimating the heart rate, we employed a data augmentation method based on a frame interpolation model. Moreover, we designed BBP-Net to infer blood pressure within a predefined range by incorporating a scaled sigmoid function. Our method resulted in estimating the heart rate with the mean absolute error (MAE) of 1.78 BPM, reducing the MAE by 34.31 % compared to the recent method, on the MMSE-HR dataset. The MAE for estimating the systolic blood pressure (SBP) and diastolic blood pressure (DBP) were 10.19 mmHg and 7.09 mmHg. On the V4V dataset, the MAE for the heart rate, SBP, and DBP were 3.83 BPM, 13.64 mmHg, and 9.4 mmHg, respectively.

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 paper proposes a two-stage deep learning framework for non-contact estimation of heart rate (HR) and blood pressure (BP) from facial video. DRP-Net extracts phase-shifted remote photoplethysmography (rPPG) signals from acral and facial regions to estimate HR (with data augmentation via frame interpolation), while BBP-Net analyzes temporal features and the phase discrepancy between these signals to regress systolic (SBP) and diastolic (DBP) blood pressure values, using a scaled sigmoid to bound outputs. On the MMSE-HR dataset it reports HR MAE of 1.78 BPM (34.31% reduction vs. a recent method), SBP MAE 10.19 mmHg and DBP MAE 7.09 mmHg; on V4V the corresponding MAEs are 3.83 BPM, 13.64 mmHg and 9.4 mmHg.

Significance. If the phase-discrepancy cue can be shown to carry BP information independent of HR correlation, the approach would offer a concrete advance in vision-based cardiovascular monitoring by avoiding contact sensors. The reported MAEs are competitive with prior rPPG work, but the absence of any ablation isolating the phase component or external validation leaves the incremental contribution unverified.

major comments (2)
  1. [BBP-Net description] Description of BBP-Net and the scaled-sigmoid bound: the central premise that phase discrepancy between acral and facial rPPG streams supplies information usable for continuous SBP/DBP regression beyond what HR alone provides is invoked to motivate the second-stage network, yet no ablation (e.g., single-stream input, time-shifted streams to null lag, or direct HR-regressor baseline for the BP task) is reported; without such a controlled comparison the end-to-end MAEs cannot be attributed to the phase cue.
  2. [Experimental results] Results on MMSE-HR and V4V: the claimed 34.31 % MAE reduction and the absolute MAEs for HR/SBP/DBP are presented without error bars, subject-wise cross-validation protocol, or re-implementation details of the baseline method, rendering the quantitative claims difficult to interpret or reproduce.
minor comments (2)
  1. [Abstract] Abstract contains the typo 'photoplethyomography' (should be photoplethysmography) and refers to the work as 'this thesis' while the submission is an arXiv paper.
  2. [Methods] Notation for the scaled sigmoid bound and the precise definition of 'phase discrepancy' feature fed to BBP-Net should be formalized with an equation rather than left descriptive.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback on our manuscript. We address the major comments point by point below, agreeing where the manuscript requires strengthening and outlining the specific revisions we will implement.

read point-by-point responses

  1. Referee: [BBP-Net description] Description of BBP-Net and the scaled-sigmoid bound: the central premise that phase discrepancy between acral and facial rPPG streams supplies information usable for continuous SBP/DBP regression beyond what HR alone provides is invoked to motivate the second-stage network, yet no ablation (e.g., single-stream input, time-shifted streams to null lag, or direct HR-regressor baseline for the BP task) is reported; without such a controlled comparison the end-to-end MAEs cannot be attributed to the phase cue.

    Authors: We agree that the absence of ablation studies isolating the contribution of the phase discrepancy is a limitation in the current manuscript. To address this, the revised version will include new experiments: (1) BBP-Net with only temporal features from a single rPPG stream, (2) a variant with artificially time-shifted streams to remove phase information, and (3) a direct baseline regressing SBP/DBP from the estimated HR alone. These controlled comparisons will quantify the incremental value of the phase cue and support the motivation for the two-stage design. revision: yes

  2. Referee: [Experimental results] Results on MMSE-HR and V4V: the claimed 34.31 % MAE reduction and the absolute MAEs for HR/SBP/DBP are presented without error bars, subject-wise cross-validation protocol, or re-implementation details of the baseline method, rendering the quantitative claims difficult to interpret or reproduce.

    Authors: We acknowledge that the experimental reporting can be improved for reproducibility and statistical rigor. In the revision we will: add error bars (standard deviations across folds or subjects) to all MAE results; provide a detailed description of the subject-wise cross-validation protocol (including number of folds and how subjects are partitioned); and clarify the source of the baseline numbers (re-implementation details or values taken from the original publication) along with any hyperparameter or preprocessing differences. revision: yes

Circularity Check

0 steps flagged

No significant circularity in claimed derivation chain.

full rationale

The paper proposes an empirical two-stage supervised deep learning pipeline (DRP-Net for dual rPPG extraction followed by BBP-Net for BP regression) and reports MAEs obtained by training and testing on the MMSE-HR and V4V datasets. No first-principles mathematical derivation, uniqueness theorem, or parameter-free prediction is asserted that reduces by construction to its own inputs or to a self-citation. The phase-discrepancy premise is introduced as an architectural design choice rather than a derived result, and the scaled-sigmoid bound is a simple output-range constraint. Standard ML evaluation on held-out splits does not constitute the enumerated circularity patterns; therefore the score is 0 and the steps list remains empty.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

Only abstract available; the approach rests on standard rPPG domain assumptions plus a large number of fitted neural-network weights and an ad-hoc bounding function whose range is chosen rather than derived.

free parameters (2)
  • DRP-Net and BBP-Net weights
    All network parameters are fitted during supervised training on video datasets.
  • BP range bounds and sigmoid scaling factors
    Predefined numeric interval and scaling constants used to constrain BBP-Net outputs.
axioms (2)
  • domain assumption Subtle skin color variations in video faithfully encode pulsatile blood volume changes (rPPG principle)
    Invoked by the definition of DRP-Net and the use of its outputs for both HR and BP.
  • domain assumption Phase and shape discrepancy between acral and facial rPPG waveforms contains usable information about arterial blood pressure
    Central modeling choice for BBP-Net; if false the second stage has no signal.

pith-pipeline@v0.9.0 · 5899 in / 1676 out tokens · 29108 ms · 2026-05-24T04:37:42.139557+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

58 extracted references · 58 canonical work pages · 1 internal anchor

  1. [1]

    Global cardiovascular diseases death rate prediction,

    O. Gaidai, Y . Cao, and S. Loginov, “Global cardiovascular diseases death rate prediction,” Current Problems in Cardiology , p. 101622, 2023. Fig. 10. Visualization of spatial attention αs. The highlighted areas illustrate regions with higher attention scores

    work page 2023

  2. [2]

    Early identification of pcos with commonly known diseases: obesity, diabetes, high blood pressure and heart disease using machine learning techniques,

    S. Aggarwal and K. Pandey, “Early identification of pcos with commonly known diseases: obesity, diabetes, high blood pressure and heart disease using machine learning techniques,” Expert Systems with Applications , vol. 217, p. 119532, 2023

    work page 2023

  3. [3]

    Heart rate variability for medical decision support systems: A review,

    O. Faust, W. Hong, H. W. Loh, S. Xu, R.-S. Tan, S. Chakraborty, P. D. Barua, F. Molinari, and U. R. Acharya, “Heart rate variability for medical decision support systems: A review,” Computers in biology and medicine, vol. 145, p. 105407, 2022

    work page 2022

  4. [4]

    Wrist pulse signal acquisition and analysis for disease diagnosis: A review,

    C. Guo, Z. Jiang, H. He, Y . Liao, and D. Zhang, “Wrist pulse signal acquisition and analysis for disease diagnosis: A review,” Computers in Biology and Medicine , vol. 143, p. 105312, 2022

    work page 2022

  5. [5]

    A support system for automatic classification of hypertension using bcg signals,

    K. Gupta, V . Bajaj, and I. A. Ansari, “A support system for automatic classification of hypertension using bcg signals,” Expert Systems with Applications, vol. 214, p. 119058, 2023

    work page 2023

  6. [6]

    Heart rate as a risk factor for cardiovascular disease,

    C. Perret-Guillaume, L. Joly, and A. Benetos, “Heart rate as a risk factor for cardiovascular disease,”Progress in cardiovascular diseases, vol. 52, no. 1, pp. 6–10, 2009

    work page 2009

  7. [7]

    High blood pressure and cardiovascular disease,

    F. D. Fuchs and P. K. Whelton, “High blood pressure and cardiovascular disease,” Hypertension, vol. 75, no. 2, pp. 285–292, 2020

    work page 2020

  8. [8]

    Pulse transit time as an indicator of arterial blood pressure,

    L. Geddes, M. V oelz, C. Babbs, J. Bourland, and W. Tacker, “Pulse transit time as an indicator of arterial blood pressure,” psychophysiology, vol. 18, no. 1, pp. 71–74, 1981

    work page 1981

  9. [9]

    Noninvasive contin- uous blood pressure estimation from pulse transit time: A review of the calibration models,

    D. Barvik, M. Cerny, M. Penhaker, and N. Noury, “Noninvasive contin- uous blood pressure estimation from pulse transit time: A review of the calibration models,” IEEE Reviews in Biomedical Engineering , vol. 15, pp. 138–151, 2021

    work page 2021

  10. [10]

    Wearable cuff-less blood pressure estimation at home via pulse transit time,

    V . G. Ganti, A. M. Carek, B. N. Nevius, J. A. Heller, M. Etemadi, and O. T. Inan, “Wearable cuff-less blood pressure estimation at home via pulse transit time,” IEEE journal of biomedical and health informatics , vol. 25, no. 6, pp. 1926–1937, 2020

    work page 1926

  11. [11]

    Deepphys: Video-based physiological mea- surement using convolutional attention networks,

    W. Chen and D. McDuff, “Deepphys: Video-based physiological mea- surement using convolutional attention networks,” in Proceedings of the european conference on computer vision (ECCV) , 2018, pp. 349–365

    work page 2018

  12. [12]

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

    Z. Yu, Y . Shen, J. Shi, H. Zhao, P. H. Torr, and G. Zhao, “Physformer: Facial video-based physiological measurement with temporal difference transformer,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 4186–4196

    work page 2022

  13. [13]

    X-ippgnet: A novel one stage deep learning architecture based on depthwise separable con- volutions for video-based pulse rate estimation,

    Y . Ouzar, D. Djeldjli, F. Bousefsaf, and C. Maaoui, “X-ippgnet: A novel one stage deep learning architecture based on depthwise separable con- volutions for video-based pulse rate estimation,” Computers in Biology and Medicine, vol. 154, p. 106592, 2023

    work page 2023

  14. [14]

    Un- supervised skin tissue segmentation for remote photoplethysmography,

    S. Bobbia, R. Macwan, Y . Benezeth, A. Mansouri, and J. Dubois, “Un- supervised skin tissue segmentation for remote photoplethysmography,” Pattern Recognition Letters, vol. 124, pp. 82–90, 2019

    work page 2019

  15. [15]

    Vipl-hr: A multi-modal database for pulse estimation from less-constrained face video,

    X. Niu, H. Han, S. Shan, and X. Chen, “Vipl-hr: A multi-modal database for pulse estimation from less-constrained face video,” in Computer Vision–ACCV 2018: 14th Asian Conference on Computer Vision, Perth, Australia, December 2–6, 2018, Revised Selected Papers, Part V 14 . Springer, 2019, pp. 562–576

    work page 2018

  16. [16]

    Multimodal spontaneous emotion corpus for human behavior analysis,

    Z. Zhang, J. M. Girard, Y . Wu, X. Zhang, P. Liu, U. Ciftci, S. Canavan, M. Reale, A. Horowitz, H. Yang et al. , “Multimodal spontaneous emotion corpus for human behavior analysis,” in Proceedings of the 13 IEEE conference on computer vision and pattern recognition , 2016, pp. 3438–3446

    work page 2016

  17. [17]

    The first vision for vitals (v4v) challenge for non-contact video-based physiological estimation,

    A. Revanur, Z. Li, U. A. Ciftci, L. Yin, and L. A. Jeni, “The first vision for vitals (v4v) challenge for non-contact video-based physiological estimation,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2021, pp. 2760–2767

    work page 2021

  18. [18]

    Dual-gan: Joint bvp and noise modeling for remote physiological measurement,

    H. Lu, H. Han, and S. K. Zhou, “Dual-gan: Joint bvp and noise modeling for remote physiological measurement,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 12 404–12 413

    work page 2021

  19. [19]

    Mspm: a multi-site physiological monitoring dataset for remote pulse, respiration, and blood pressure estimation,

    J. Speth, N. Vance, B. Sporrer, L. Niu, P. Flynn, and A. Czajka, “Mspm: a multi-site physiological monitoring dataset for remote pulse, respiration, and blood pressure estimation,” IEEE Transactions on In- strumentation and Measurement , 2024

    work page 2024

  20. [20]

    Realistic pulse waveforms estimation via contrastive learning in remote photoplethysmography,

    B. Dong, Y . Liu, K. Yang, and J. Cao, “Realistic pulse waveforms estimation via contrastive learning in remote photoplethysmography,” IEEE Transactions on Instrumentation and Measurement , 2024

    work page 2024

  21. [21]

    Camera-based blood pressure monitoring based on multi-site and multi-wavelength pulse transit time features,

    Y . Huang, D. Huang, J. Huang, G. Wang, L. Pan, H. Lu, M. He, and W. Wang, “Camera-based blood pressure monitoring based on multi-site and multi-wavelength pulse transit time features,” IEEE Transactions on Instrumentation and Measurement , 2024

    work page 2024

  22. [22]

    Exploiting multi-wavelength morphological features of camera-ppg for blood pressure estimation,

    Z. Hou, D. Huang, Y . Huang, J. Huang, N. Zhao, L. Pan, H. Lu, C. Shan, and W. Wang, “Exploiting multi-wavelength morphological features of camera-ppg for blood pressure estimation,” IEEE Transactions on Instrumentation and Measurement , 2025

    work page 2025

  23. [23]

    Advancements in non- contact, multiparameter physiological measurements using a webcam,

    M.-Z. Poh, D. J. McDuff, and R. W. Picard, “Advancements in non- contact, multiparameter physiological measurements using a webcam,” IEEE transactions on biomedical engineering , vol. 58, no. 1, pp. 7–11, 2010

    work page 2010

  24. [24]

    Measuring pulse rate with a webcam—a non-contact method for evaluating car- diac activity,

    M. Lewandowska, J. Rumi ´nski, T. Kocejko, and J. Nowak, “Measuring pulse rate with a webcam—a non-contact method for evaluating car- diac activity,” in 2011 federated conference on computer science and information systems (FedCSIS). IEEE, 2011, pp. 405–410

    work page 2011

  25. [25]

    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

  26. [26]

    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

  27. [27]

    The benefit of distraction: Denoising camera-based physiological measurements using inverse attention,

    E. M. Nowara, D. McDuff, and A. Veeraraghavan, “The benefit of distraction: Denoising camera-based physiological measurements using inverse attention,” in Proceedings of the IEEE/CVF international con- ference on computer vision , 2021, pp. 4955–4964

    work page 2021

  28. [28]

    arXiv preprint arXiv:1905.02419 , year =

    Z. Yu, X. Li, and G. Zhao, “Remote photoplethysmograph signal measurement from facial videos using spatio-temporal networks,” arXiv preprint arXiv:1905.02419, 2019

    work page arXiv 1905

  29. [29]

    Physformer++: Facial video-based physiological measurement with slowfast temporal difference transformer,

    Z. Yu, Y . Shen, J. Shi, H. Zhao, Y . Cui, J. Zhang, P. Torr, and G. Zhao, “Physformer++: Facial video-based physiological measurement with slowfast temporal difference transformer,” International Journal of Computer Vision, vol. 131, no. 6, pp. 1307–1330, 2023

    work page 2023

  30. [30]

    Continuous blood pressure measurement from one-channel electrocardiogram signal using deep-learning techniques,

    F. Miao, B. Wen, Z. Hu, G. Fortino, X.-P. Wang, Z.-D. Liu, M. Tang, and Y . Li, “Continuous blood pressure measurement from one-channel electrocardiogram signal using deep-learning techniques,” Artificial In- telligence in Medicine , vol. 108, p. 101919, 2020

    work page 2020

  31. [31]

    Pp-net: A deep learning framework for ppg-based blood pressure and heart rate estimation,

    M. Panwar, A. Gautam, D. Biswas, and A. Acharyya, “Pp-net: A deep learning framework for ppg-based blood pressure and heart rate estimation,” IEEE Sensors Journal , vol. 20, no. 17, pp. 10 000–10 011, 2020

    work page 2020

  32. [32]

    Mlp-bp: A novel framework for cuffless blood pressure measurement with ppg and ecg signals based on mlp-mixer neural networks,

    B. Huang, W. Chen, C.-L. Lin, C.-F. Juang, and J. Wang, “Mlp-bp: A novel framework for cuffless blood pressure measurement with ppg and ecg signals based on mlp-mixer neural networks,” Biomedical Signal Processing and Control, vol. 73, p. 103404, 2022

    work page 2022

  33. [33]

    Stp: Self-supervised transfer learning based on transformer for noninvasive blood pressure estimation using photoplethysmography,

    C. Ma, P. Zhang, H. Zhang, Z. Liu, F. Song, Y . He, and G. Zhang, “Stp: Self-supervised transfer learning based on transformer for noninvasive blood pressure estimation using photoplethysmography,” Expert Systems with Applications, vol. 249, p. 123809, 2024

    work page 2024

  34. [34]

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

    B.-F. Wu, B.-J. Wu, B.-R. Tsai, and C.-P. 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

  35. [35]

    Estimation of blood pressure waveform from facial video using a deep u-shaped network and the wavelet representation of imaging photoplethysmographic signals,

    F. Bousefsaf, T. Desquins, D. Djeldjli, Y . Ouzar, C. Maaoui, and A. Pruski, “Estimation of blood pressure waveform from facial video using a deep u-shaped network and the wavelet representation of imaging photoplethysmographic signals,” Biomedical Signal Processing and Control, vol. 78, p. 103895, 2022

    work page 2022

  36. [36]

    Remote blood pressure estimation via the spatiotemporal mapping of facial videos,

    Y . Chen, J. Zhuang, B. Li, Y . Zhang, and X. Zheng, “Remote blood pressure estimation via the spatiotemporal mapping of facial videos,” Sensors, vol. 23, no. 6, p. 2963, 2023

    work page 2023

  37. [37]

    Joint face detection and alignment using multitask cascaded convolutional networks,

    K. Zhang, Z. Zhang, Z. Li, and Y . Qiao, “Joint face detection and alignment using multitask cascaded convolutional networks,” IEEE signal processing letters , vol. 23, no. 10, pp. 1499–1503, 2016

    work page 2016

  38. [38]

    An advanced detrending method with application to hrv analysis,

    M. P. Tarvainen, P. O. Ranta-Aho, and P. A. Karjalainen, “An advanced detrending method with application to hrv analysis,” IEEE transactions on biomedical engineering , vol. 49, no. 2, pp. 172–175, 2002

    work page 2002

  39. [39]

    A. C. Guyton, Text book of medical physiology . China, 2006

    work page 2006

  40. [40]

    Film: Frame interpolation for large motion,

    F. Reda, J. Kontkanen, E. Tabellion, D. Sun, C. Pantofaru, and B. Cur- less, “Film: Frame interpolation for large motion,” in European Confer- ence on Computer Vision . Springer, 2022, pp. 250–266

    work page 2022

  41. [41]

    Autohr: A strong end-to- end baseline for remote heart rate measurement with neural searching,

    Z. Yu, X. Li, X. Niu, J. Shi, and G. Zhao, “Autohr: A strong end-to- end baseline for remote heart rate measurement with neural searching,” IEEE Signal Processing Letters , vol. 27, pp. 1245–1249, 2020

    work page 2020

  42. [42]

    And-rppg: A novel denoising-rppg network for improving remote heart rate estimation,

    B. Lokendra and G. Puneet, “And-rppg: A novel denoising-rppg network for improving remote heart rate estimation,” Computers in biology and medicine, vol. 141, p. 105146, 2022

    work page 2022

  43. [43]

    Facial video-based remote physiolog- ical measurement via self-supervised learning,

    Z. Yue, M. Shi, and S. Ding, “Facial video-based remote physiolog- ical measurement via self-supervised learning,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2023

    work page 2023

  44. [44]

    BAM: Bottleneck Attention Module

    J. Park, S. Woo, J.-Y . Lee, and I. S. Kweon, “Bam: Bottleneck attention module,” arXiv preprint arXiv:1807.06514 , 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  45. [45]

    Ppg-based blood pressure estimation can benefit from scalable multi-scale fusion neural networks and multi- task learning,

    Q. Hu, D. Wang, and C. Yang, “Ppg-based blood pressure estimation can benefit from scalable multi-scale fusion neural networks and multi- task learning,” Biomedical Signal Processing and Control , vol. 78, p. 103891, 2022

    work page 2022

  46. [46]

    Robust estimation of a location parameter in the presence of asymmetry,

    J. R. Collins, “Robust estimation of a location parameter in the presence of asymmetry,” The Annals of Statistics , pp. 68–85, 1976

    work page 1976

  47. [47]

    Heart rate estimation network from facial videos using spatiotemporal feature image,

    K. B. Jaiswal and T. Meenpal, “Heart rate estimation network from facial videos using spatiotemporal feature image,” Computers in Biology and Medicine, vol. 151, p. 106307, 2022

    work page 2022

  48. [48]

    Efficientphys: Enabling simple, fast and accurate camera-based vitals measurement,

    X. Liu, B. L. Hill, Z. Jiang, S. Patel, and D. McDuff, “Efficientphys: Enabling simple, fast and accurate camera-based vitals measurement,” arXiv preprint arXiv:2110.04447 , 2021

    work page arXiv 2021

  49. [49]

    Using high-fidelity avatars to advance camera-based cardiac pulse measure- ment,

    D. McDuff, J. Hernandez, X. Liu, E. Wood, and T. Baltrusaitis, “Using high-fidelity avatars to advance camera-based cardiac pulse measure- ment,” IEEE Transactions on Biomedical Engineering , vol. 69, no. 8, pp. 2646–2656, 2022

    work page 2022

  50. [50]

    Channel-wise interactive learning for remote heart rate estimation from facial video,

    Q. Li, D. Guo, W. Qian, X. Tian, X. Sun, H. Zhao, and M. Wang, “Channel-wise interactive learning for remote heart rate estimation from facial video,” IEEE Transactions on Circuits and Systems for Video Technology, 2023

    work page 2023

  51. [51]

    Dual-path tokenlearner for remote photoplethysmography-based phys- iological measurement with facial videos,

    W. Qian, D. Guo, K. Li, X. Zhang, X. Tian, X. Yang, and M. Wang, “Dual-path tokenlearner for remote photoplethysmography-based phys- iological measurement with facial videos,” IEEE Transactions on Com- putational Social Systems , 2024

    work page 2024

  52. [52]

    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, p. 102328, 2021

    work page 2021

  53. [53]

    As- sessment of non-invasive blood pressure prediction from ppg and rppg signals using deep learning,

    F. Schrumpf, P. Frenzel, C. Aust, G. Osterhoff, and M. Fuchs, “As- sessment of non-invasive blood pressure prediction from ppg and rppg signals using deep learning,” Sensors, vol. 21, no. 18, p. 6022, 2021

    work page 2021

  54. [54]

    A study of projection- based attentive spatial–temporal map for remote photoplethysmography measurement,

    D.-Y . Kim, S.-Y . Cho, K. Lee, and C.-B. Sohn, “A study of projection- based attentive spatial–temporal map for remote photoplethysmography measurement,” Bioengineering, vol. 9, no. 11, p. 638, 2022

    work page 2022

  55. [55]

    Neural network model combination for video-based blood pressure estimation: New approach and evaluation,

    B. Hamoud, A. Kashevnik, W. Othman, and N. Shilov, “Neural network model combination for video-based blood pressure estimation: New approach and evaluation,” Sensors, vol. 23, no. 4, p. 1753, 2023

    work page 2023

  56. [56]

    The validity and practicality of sun-reactive skin types i through vi,

    T. B. Fitzpatrick, “The validity and practicality of sun-reactive skin types i through vi,” Archives of dermatology , vol. 124, no. 6, pp. 869–871, 1988

    work page 1988

  57. [57]

    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,” Ad- vances in Neural Information Processing Systems , vol. 33, pp. 19 400– 19 411, 2020

    work page 2020

  58. [58]

    The british hypertension society protocol for the evaluation of automated and semi- automated blood pressure measuring devices with special reference to ambulatory systems,

    E. O’Brien, J. Petrie, W. Littler, M. de Swiet, P. L. Padfield, K. O’Malley, M. Jamieson, D. Altman, M. Bland, and N. Atkins, “The british hypertension society protocol for the evaluation of automated and semi- automated blood pressure measuring devices with special reference to ambulatory systems,” Journal of hypertension , vol. 8, no. 7, pp. 607– 619, 1990

    work page 1990