pith. sign in

arxiv: 2605.23174 · v1 · pith:I7JWEX4Rnew · submitted 2026-05-22 · 💻 cs.CV

LQ-rPPG: A Label-Quantized Coarse-to-Fine Learning Framework for Remote Physiological Measurement

Jun Seong Lee , Samyeul Noh , Changki Sung , Hyun Myung This is my paper

Pith reviewed 2026-05-25 05:11 UTC · model grok-4.3

classification 💻 cs.CV
keywords remote photoplethysmographyrPPGlabel quantizationcoarse-to-fine learningphysiological measurementdeep learningnoise reduction
0
0 comments X

The pith

Quantizing noisy contact PPG labels into multi-bit pseudo labels enables a coarse-to-fine model to learn robust rPPG signals from facial videos.

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

The paper shows that contact PPG signals used as training labels contain noise from motion and sensor issues, causing deep models to overfit and generalize poorly. To fix this, continuous labels are turned into multi-bit quantized pseudo labels that keep physiological content but drop variability. These cleaner labels then provide hierarchical supervision to a coarse-to-fine network that refines the rPPG estimate step by step. The result is stronger intra- and cross-dataset performance on benchmarks together with an 88 percent drop in parameters and 191 percent higher throughput.

Core claim

LQ-rPPG consists of a label quantization module and a coarse-to-fine rPPG estimation model. The label quantization module transforms continuous PPG signals into multi-bit quantized pseudo labels with reduced noise and variability. The coarse-to-fine estimation model progressively refines rPPG signals under hierarchical supervision guided by the multi-bit pseudo labels. This design alleviates overfitting to label-specific variations and enables the model to learn structured and consistent representations, achieving robust and generalizable rPPG estimation even under challenging conditions.

What carries the argument

Label quantization module that produces multi-bit pseudo labels for hierarchical supervision inside a coarse-to-fine rPPG estimation model.

If this is right

  • Strong performance in both intra- and cross-dataset evaluations on multiple benchmark datasets.
  • Parameters reduced by 88 percent and multiply-accumulate operations reduced by 29 percent.
  • Throughput increased by 191 percent.
  • Robust rPPG estimation remains possible under challenging motion and lighting conditions.

Where Pith is reading between the lines

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

  • The same quantization-plus-hierarchical-supervision pattern could be tested on other noisy physiological signals such as ECG or respiration rate from video.
  • The reported model compression makes real-time rPPG feasible on mobile or embedded hardware for continuous monitoring.
  • Cross-dataset gains imply the learned features are more invariant to individual recording setups than earlier end-to-end networks.

Load-bearing premise

The quantization step removes noise from PPG labels without discarding the underlying physiological information needed for accurate rPPG learning.

What would settle it

A controlled experiment showing that a model trained on the original noisy PPG labels outperforms the quantized-label version on cross-dataset tests would falsify the claim.

Figures

Figures reproduced from arXiv: 2605.23174 by Changki Sung, Hyun Myung, Jun Seong Lee, Samyeul Noh.

Figure 1
Figure 1. Figure 1: Examples of noise and variability in contact-based PPG signals used as ground [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of LQ-rPPG. The framework consists of (a) a label quantization module [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Label quantization module. The module transforms a continuous PPG signal into [PITH_FULL_IMAGE:figures/full_fig_p015_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Coarse-to-fine rPPG estimation model. The model estimates rPPG signals from [PITH_FULL_IMAGE:figures/full_fig_p020_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Visualization of results on the PURE, UBFC, V4V, and MMPD datasets. [PITH_FULL_IMAGE:figures/full_fig_p030_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Comparison of intra-dataset testing results on the MMPD dataset under six super [PITH_FULL_IMAGE:figures/full_fig_p037_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Codebook utilization across different bit levels on the UBFC dataset. [PITH_FULL_IMAGE:figures/full_fig_p039_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Effect of the classification loss coefficient [PITH_FULL_IMAGE:figures/full_fig_p044_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Comparison of C2F model training loss curves under the two-stage setting and the [PITH_FULL_IMAGE:figures/full_fig_p046_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Qualitative failure cases in the MMPD walking scenario. The estimated rPPG [PITH_FULL_IMAGE:figures/full_fig_p049_10.png] view at source ↗
read the original abstract

Remote photoplethysmography (rPPG) enables non-contact measurement of physiological signals from facial videos, offering strong potential for remote healthcare and daily health monitoring. Driven by this potential, various deep learning-based rPPG methods have been proposed to improve rPPG estimation. However, previous deep learning-based rPPG methods have paid little attention to the quality of training labels and their impact on model learning. Contact-based PPG signals used as training labels often contain noise and variability caused by motion artifacts, inconsistent sensor contact, and morphological distortions. Such label inconsistency can lead models to overfit to the label noise and variability and consequently degrade generalization performance. To address this issue, we propose LQ-rPPG, a label-quantized coarse-to-fine learning framework for robust rPPG estimation. LQ-rPPG consists of a label quantization module and a coarse-to-fine rPPG estimation model. The label quantization module transforms continuous PPG signals into multi-bit quantized pseudo labels with reduced noise and variability. The coarse-to-fine estimation model progressively refines rPPG signals under hierarchical supervision guided by the multi-bit pseudo labels. This design alleviates overfitting to label-specific variations and enables the model to learn structured and consistent representations. As a result, LQ-rPPG achieves robust and generalizable rPPG estimation even under challenging conditions. Experiments on multiple benchmark datasets demonstrate that LQ-rPPG achieves strong performance in both intra- and cross-dataset evaluations, while reducing parameters and multiply-accumulate operations by 88% and 29%, respectively, and increasing throughput by 191%. The code is available at https://github.com/Anonymous-repo-code/LQ-rPPG.

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 / 0 minor

Summary. The manuscript presents LQ-rPPG, a label-quantized coarse-to-fine learning framework for remote photoplethysmography (rPPG). It consists of a label quantization module that maps noisy continuous PPG signals to multi-bit quantized pseudo labels and a coarse-to-fine estimation model trained under hierarchical supervision from those labels. The central claim is that this design reduces overfitting to label noise/variability, yielding robust intra- and cross-dataset performance on benchmark rPPG datasets while cutting parameters by 88%, MACs by 29%, and raising throughput by 191%. Code is released at the cited GitHub repository.

Significance. If the quantization step is shown to preserve the necessary frequency content and amplitude dynamics, the work would address a recognized practical bottleneck in supervised rPPG learning (label inconsistency from contact sensors). The reported efficiency numbers, if reproducible, would be practically useful for deployment. Public code is a clear strength that enables verification and extension.

major comments (2)
  1. [Abstract] Abstract: the claim that multi-bit quantized pseudo labels 'reduce noise and variability' while 'preserving the physiological information needed for effective hierarchical supervision' is asserted without any frequency-domain comparison, amplitude-distribution analysis, or ablation that isolates quantization bit-width from the rest of the pipeline. This is load-bearing because the coarse-to-fine hierarchy is explicitly supervised by these pseudo labels; if quantization acts as an unintended low-pass filter or clips peak-to-peak variation, the reported cross-dataset gains could be dataset artifacts rather than a general solution.
  2. [Abstract / Experiments] Abstract / Experiments section: no description of data splits, subject-wise partitioning, or ablation tables isolating the quantization module is referenced, so it is impossible to determine whether the stated performance and efficiency gains are attributable to the proposed components or to implementation details and particular dataset characteristics.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful and constructive comments. We address each major point below and agree that targeted additions will strengthen the manuscript's clarity and evidentiary support. Revisions will be made accordingly.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that multi-bit quantized pseudo labels 'reduce noise and variability' while 'preserving the physiological information needed for effective hierarchical supervision' is asserted without any frequency-domain comparison, amplitude-distribution analysis, or ablation that isolates quantization bit-width from the rest of the pipeline. This is load-bearing because the coarse-to-fine hierarchy is explicitly supervised by these pseudo labels; if quantization acts as an unintended low-pass filter or clips peak-to-peak variation, the reported cross-dataset gains could be dataset artifacts rather than a general solution.

    Authors: We agree that the abstract claim would be more robust with direct supporting analyses. The manuscript contains bit-width ablations (Section 4.4) and qualitative signal visualizations (Figure 4), but lacks explicit frequency-domain or amplitude-distribution comparisons. In revision we will add a dedicated analysis subsection with PSD plots of original vs. quantized labels across bit-widths, peak-to-peak amplitude statistics, and an expanded ablation that isolates the quantization module from the coarse-to-fine hierarchy. These additions will be referenced from the abstract. revision: yes

  2. Referee: [Abstract / Experiments] Abstract / Experiments section: no description of data splits, subject-wise partitioning, or ablation tables isolating the quantization module is referenced, so it is impossible to determine whether the stated performance and efficiency gains are attributable to the proposed components or to implementation details and particular dataset characteristics.

    Authors: The experiments section specifies the four benchmark datasets and follows standard subject-independent protocols (e.g., leave-one-subject-out or non-overlapping subject partitions for train/test). Ablation tables that isolate the quantization module appear in Table 2. To address the referee's concern about explicit referencing, we will insert a concise description of the partitioning strategy and direct citations to the relevant ablation tables into both the abstract and the opening paragraph of the experiments section. revision: yes

Circularity Check

0 steps flagged

No circularity; proposed architecture is forward and empirically validated

full rationale

The paper introduces a new label-quantization module and coarse-to-fine model as design choices to address label noise in rPPG training. No equations or claims reduce a result to a fitted parameter or self-citation by construction. Performance claims rest on benchmark experiments rather than definitional equivalence. The quantization step is presented as an input transformation whose benefit is measured externally, not presupposed.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 1 invented entities

Abstract-only review provides no equations, implementation details, or hyperparameter lists; therefore free parameters, axioms, and invented entities cannot be enumerated beyond the high-level description of the quantization step.

invented entities (1)
  • multi-bit quantized pseudo labels no independent evidence
    purpose: Reduce noise and variability in contact PPG training labels
    Introduced by the label quantization module; no independent evidence or validation details given in abstract.

pith-pipeline@v0.9.0 · 5858 in / 1085 out tokens · 35469 ms · 2026-05-25T05:11:25.468140+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

67 extracted references · 67 canonical work pages · 3 internal anchors

  1. [1]

    Optics Express , volume =

    Remote plethysmographic imaging using ambient light , author =. Optics Express , volume =. 2008 , doi =

    work page 2008

  2. [2]

    IEEE Transactions on Instrumentation and Measurement , volume =

    Video-based heart rate measurement: Recent advances and future prospects , author =. IEEE Transactions on Instrumentation and Measurement , volume =. 2019 , doi =

    work page 2019

  3. [3]

    IEEE Signal Processing Magazine , volume =

    Facial-video-based physiological signal measurement: Recent advances and affective applications , author =. IEEE Signal Processing Magazine , volume =. 2021 , doi =

    work page 2021

  4. [4]

    Expert Systems with Applications , volume =

    Trusted emotion recognition based on multiple signals captured from video , author =. Expert Systems with Applications , volume =. 2023 , doi =

    work page 2023

  5. [5]

    Expert Systems with Applications , volume =

    A novel temporal adaptive fuzzy neural network for facial feature based fatigue assessment , author =. Expert Systems with Applications , volume =. 2024 , doi =

    work page 2024

  6. [6]

    Medical Engineering & Physics , volume =

    Heart rate measurement based on a time-lapse image , author =. Medical Engineering & Physics , volume =. 2007 , doi =

    work page 2007

  7. [7]

    Sensors , volume =

    Evaluation of remote photoplethysmography measurement conditions toward telemedicine applications , author =. Sensors , volume =. 2021 , doi =

    work page 2021

  8. [8]

    arXiv preprint arXiv:2506.09718 , year=

    Non-contact health monitoring during daily personal care routines , author=. arXiv preprint arXiv:2506.09718 , year=

    work page arXiv

  9. [9]

    Biomedical Engineering Online , volume =

    A comprehensive review of heart rate measurement using remote photoplethysmography and deep learning , author =. Biomedical Engineering Online , volume =. 2025 , doi =

    work page 2025

  10. [10]

    Proceedings of the British Machine Vision Conference , pages=

    Visual heart rate estimation with convolutional neural network , author=. Proceedings of the British Machine Vision Conference , pages=

    work page

  11. [11]

    2018 , doi =

    Chen, Weixuan and McDuff, Daniel , booktitle =. 2018 , doi =

    work page 2018

  12. [12]

    Lee, Eugene and Chen, Evan and Lee, Chen-Yi , booktitle =. Meta-. 2020 , doi =

    work page 2020

  13. [13]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages =

    The benefit of distraction: denoising camera-based physiological measurements using inverse attention , author =. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages =. 2021 , doi =

    work page 2021

  14. [14]

    2019 , doi =

    Niu, Xuesong and Shan, Shiguang and Han, Hu and Chen, Xilin , journal =. 2019 , doi =

    work page 2019

  15. [15]

    Proceedings of the European Conference on Computer Vision , pages =

    Video-based remote physiological measurement via cross-verified feature disentangling , author =. Proceedings of the European Conference on Computer Vision , pages =. 2020 , doi =

    work page 2020

  16. [16]

    arXiv preprint arXiv:1905.02419 , year =

    Remote photoplethysmograph signal measurement from facial videos using spatio-temporal networks , author =. arXiv preprint arXiv:1905.02419 , year =

    work page arXiv 1905

  17. [17]

    2022 , doi =

    Botina-Monsalve, Deivid and Benezeth, Yannick and Miteran, Johel , booktitle =. 2022 , doi =

    work page 2022

  18. [18]

    2023 , doi =

    Lee, Jun Seong and Hwang, Gyutae and Ryu, Moonwook and Lee, Sang Jun , booktitle =. 2023 , doi =

    work page 2023

  19. [19]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume =

    Learning motion-robust remote photoplethysmography through arbitrary resolution videos , author =. Proceedings of the AAAI Conference on Artificial Intelligence , volume =. 2023 , doi =

    work page 2023

  20. [20]

    2022 , doi =

    Yu, Zitong and Shen, Yuming and Shi, Jingang and Zhao, Hengshuang and Torr, Philip HS and Zhao, Guoying , booktitle=. 2022 , doi =

    work page 2022

  21. [21]

    2023 , doi =

    Yu, Zitong and Shen, Yuming and Shi, Jingang and Zhao, Hengshuang and Cui, Yawen and Zhang, Jiehua and Torr, Philip and Zhao, Guoying , journal=. 2023 , doi =

    work page 2023

  22. [22]

    2023 , doi =

    Shao, Hang and Luo, Lei and Qian, Jianjun and Chen, Shuo and Hu, Chuanfei and Yang, Jian , journal=. 2023 , doi =

    work page 2023

  23. [23]

    2024 , doi =

    Choi, Jae-Ho and Kang, Ki-Bong and Kim, Kyung-Tae , booktitle=. 2024 , doi =

    work page 2024

  24. [24]

    2025 , doi =

    Zou, Bochao and Guo, Zizheng and Chen, Jiansheng and Zhuo, Junbao and Huang, Weiran and Ma, Huimin , journal=. 2025 , doi =

    work page 2025

  25. [25]

    2024 , doi =

    Luo, Chaoqi and Xie, Yiping and Yu, Zitong , booktitle=. 2024 , doi =

    work page 2024

  26. [26]

    2025 , doi =

    Zou, Bochao and Guo, Zizheng and Hu, Xiaocheng and Ma, Huimin , booktitle=. 2025 , doi =

    work page 2025

  27. [27]

    2021 , doi =

    Bhat, Shariq Farooq and Alhashim, Ibraheem and Wonka, Peter , booktitle=. 2021 , doi =

    work page 2021

  28. [28]

    2022 , doi =

    Bhat, Shariq Farooq and Alhashim, Ibraheem and Wonka, Peter , booktitle=. 2022 , doi =

    work page 2022

  29. [29]

    Shao, Shuwei and Pei, Zhongcai and Wu, Xingming and Liu, Zhong and Chen, Weihai and Li, Zhengguo , journal=

    work page

  30. [30]

    2022 , doi =

    Li, Yanjie and Yang, Sen and Liu, Peidong and Zhang, Shoukui and Wang, Yunxiao and Wang, Zhicheng and Yang, Wankou and Xia, Shu-Tao , booktitle=. 2022 , doi =

    work page 2022

  31. [31]

    Proceedings of the European Conference on Computer Vision , pages=

    Arbitrary-oriented object detection with circular smooth label , author=. Proceedings of the European Conference on Computer Vision , pages=. 2020 , doi =

    work page 2020

  32. [32]

    Journal of Biomedical Optics , volume =

    Motion-compensated noncontact imaging photoplethysmography to monitor cardiorespiratory status during exercise , author =. Journal of Biomedical Optics , volume =. 2011 , doi =

    work page 2011

  33. [33]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =

    Remote heart rate measurement from face videos under realistic situations , author =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =. 2014 , doi =

    work page 2014

  34. [34]

    Feng, Litong and Po, Lai Man and Xu, Xuyuan and Li, Yueting and Cheung, Tak Ming , booktitle =. Dynamic. 2015 , doi =

    work page 2015

  35. [35]

    2016 23rd International Conference on Pattern Recognition , pages =

    Remote photoplethysmography based on implicit living skin tissue segmentation , author =. 2016 23rd International Conference on Pattern Recognition , pages =. 2016 , doi =

    work page 2016

  36. [36]

    IEEE Transactions on Biomedical Engineering , volume =

    Advancements in noncontact, multiparameter physiological measurements using a webcam , author =. IEEE Transactions on Biomedical Engineering , volume =. 2010 , doi =

    work page 2010

  37. [37]

    Optics Express , volume =

    Non-contact, automated cardiac pulse measurements using video imaging and blind source separation , author =. Optics Express , volume =. 2010 , doi =

    work page 2010

  38. [38]

    IEEE Transactions on Biomedical Engineering , volume =

    Improvements in remote cardiopulmonary measurement using a five band digital camera , author =. IEEE Transactions on Biomedical Engineering , volume =. 2014 , doi =

    work page 2014

  39. [39]

    Exploiting spatial redundancy of image sensor for motion robust

    Wang, Wenjin and Stuijk, Sander and De Haan, Gerard , journal =. Exploiting spatial redundancy of image sensor for motion robust. 2014 , doi =

    work page 2014

  40. [40]

    Robust pulse rate from chrominance-based

    De Haan, Gerard and Jeanne, Vincent , journal =. Robust pulse rate from chrominance-based. 2013 , doi =

    work page 2013

  41. [41]

    Improved motion robustness of remote-

    De Haan, Gerard and Van Leest, Arno , journal =. Improved motion robustness of remote-. 2014 , doi =

    work page 2014

  42. [42]

    Algorithmic principles of remote

    Wang, Wenjin and Den Brinker, Albertus Christiaan and Stuijk, Sander and De Haan, Gerard , journal =. Algorithmic principles of remote. 2016 , doi =

    work page 2016

  43. [43]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops , pages =

    Non-contact heart rate monitoring by combining convolutional neural network skin detection and remote photoplethysmography via a low-cost camera , author =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops , pages =. 2018 , doi =

    work page 2018

  44. [44]

    2025 , doi =

    Liu, Tianqi and Xiao, Hanguang and Sun, Yisha and Li, Yulin and Zhao, Shiyi and Yi, Zhenyu and Zhao, Aohui , journal=. 2025 , doi =

    work page 2025

  45. [45]

    2025 , doi =

    Li, Jiajie and Cheng, Juan and Song, Rencheng and Liu, Yu , journal=. 2025 , doi =

    work page 2025

  46. [46]

    Zhu, Lianghui and Liao, Bencheng and Zhang, Qian and Wang, Xinlong and Liu, Wenyu and Wang, Xinggang , journal=. Vision

    work page

  47. [47]

    Liu, Xin and Narayanswamy, Girish and Paruchuri, Akshay and Zhang, Xiaoyu and Tang, Jiankai and Zhang, Yuzhe and Sengupta, Roni and Patel, Shwetak and Wang, Yuntao and McDuff, Daniel , journal=

    work page

  48. [48]

    Advances in Neural Information Processing Systems , volume=

    Neural discrete representation learning , author=. Advances in Neural Information Processing Systems , volume=

    work page

  49. [49]

    Proceedings of the IEEE International Symposium on Robot and Human Interactive Communication , pages=

    Non-contact video-based pulse rate measurement on a mobile service robot , author=. Proceedings of the IEEE International Symposium on Robot and Human Interactive Communication , pages=. 2014 , doi =

    work page 2014

  50. [50]

    Pattern Recognition Letters , volume=

    Unsupervised skin tissue segmentation for remote photoplethysmography , author=. Pattern Recognition Letters , volume=. 2019 , doi =

    work page 2019

  51. [51]

    A Reproducible Study on Remote Heart Rate Measurement

    A reproducible study on remote heart rate measurement , author=. arXiv preprint arXiv:1709.00962 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  52. [52]

    and Yin, Lijun and Jeni, L

    Revanur, Ambareesh and Li, Zhihua and Ciftci, Umur A. and Yin, Lijun and Jeni, L. The first vision for vitals (. Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops , pages=

    work page

  53. [53]

    Tang, Jiankai and Chen, Kequan and Wang, Yuntao and Shi, Yuanchun and Patel, Shwetak and McDuff, Daniel and Liu, Xin , booktitle=

    work page

  54. [54]

    Decoupled Weight Decay Regularization

    Decoupled weight decay regularization , author=. arXiv preprint arXiv:1711.05101 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  55. [55]

    A disciplined approach to neural network hyper-parameters: Part 1 -- learning rate, batch size, momentum, and weight decay

    A disciplined approach to neural network hyper-parameters: Part 1 -- learning rate, batch size, momentum, and weight decay , author=. arXiv preprint arXiv:1803.09820 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  56. [56]

    Advances in Neural Information Processing Systems , volume=

    Multi-task temporal shift attention networks for on-device contactless vitals measurement , author=. Advances in Neural Information Processing Systems , volume=

    work page

  57. [57]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    The way to my heart is through contrastive learning: Remote photoplethysmography from unlabelled video , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    work page

  58. [58]

    2021 , doi =

    Lu, Hao and Han, Hu and Zhou, S Kevin , booktitle=. 2021 , doi =

    work page 2021

  59. [59]

    2023 , doi =

    Liu, Xin and Hill, Brian and Jiang, Ziheng and Patel, Shwetak and McDuff, Daniel , booktitle=. 2023 , doi =

    work page 2023

  60. [60]

    2024 , doi=

    Sun, Zhaodong and Li, Xiaobai , journal=. 2024 , doi=

    work page 2024

  61. [61]

    IEEE Journal of Biomedical and Health Informatics , volume=

    Chu, Shuyang and Xia, Menghan and Yuan, Mengyao and Liu, Xin and Sepp. IEEE Journal of Biomedical and Health Informatics , volume=. 2025 , doi =

    work page 2025

  62. [62]

    Spiking-

    Liu, Mingxuan and Tang, Jiankai and Chen, Yongli and Li, Haoxiang and Qi, Jiahao and Li, Siwei and Wang, Kegang and Gan, Jie and Wang, Yuntao and Chen, Hong , journal=. Spiking-. 2025 , doi =

    work page 2025

  63. [63]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Contactless pulse estimation leveraging pseudo labels and self-supervision , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    work page

  64. [64]

    2020 , doi =

    Yu, Zitong and Li, Xiaobai and Niu, Xuesong and Shi, Jingang and Zhao, Guoying , journal=. 2020 , doi =

    work page 2020

  65. [65]

    2026 , howpublished =

    Official Implementation of LQ-rPPG , author =. 2026 , howpublished =

    work page 2026

  66. [66]

    Sun, Zhaodong and Li, Xiaobai , booktitle=

    work page

  67. [67]

    Behavior research methods , volume=

    Makowski, Dominique and Pham, Tam and Lau, Zen J and Brammer, Jan C and Lespinasse, Fran. Behavior research methods , volume=

    work page