pith. sign in

arxiv: 2606.28089 · v1 · pith:ZMAA4PHHnew · submitted 2026-06-26 · 💻 cs.CV

RPM-Distill: Physiology-guided Adaptive Cross-modal Distillation for Robust Remote Physiological Measurement

Jiyao Wang , Qingyong Hu , Duoxun Tang , Xiao Yang , Kaishun Wu , Jiangbo Yu This is my paper

Pith reviewed 2026-06-29 04:14 UTC · model grok-4.3

classification 💻 cs.CV
keywords remote physiological measurementcross-modal distillationradar sensingvideo-based monitoringfrequency domain analysisadaptive distillationheart rate estimationphysiological signal processing
0
0 comments X

The pith

Distilling radar frequency rhythms into video models during training yields more robust remote heart-rate and breathing estimates from video alone.

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

The paper establishes that video-based remote physiological measurement can be made reliable under varying light, skin tone, and motion by transferring knowledge from radar only while the model is trained. Although RGB video and RF radar differ in how they sense the body and look in the time domain, they share latent periodic patterns in the frequency domain that carry the cardiac and respiratory signals. The method anchors the main frequency peak, aligns the surrounding spectrum, and keeps overall shape while an adaptive network decides per sample how much to trust the radar teacher, avoiding harmful transfers. If this holds, video systems become accurate enough for everyday use without needing radar hardware at deployment. Experiments across hard conditions and different datasets confirm the gains over video-only baselines.

Core claim

RPM-Distill performs physiology-guided cross-modal distillation that transfers spectral evidence of periodic rhythms from synchronized radar to a video student via three losses on peak anchoring, background distribution matching, and morphology preservation, with a spectral policy network that predicts sample-specific gates and weights from the student-teacher relation map under a meta bilevel objective so that video-only inference remains robust.

What carries the argument

The spectral policy network that generates per-sample distillation gates and loss-component weights from the relation map between student and teacher spectra.

If this is right

  • Video models achieve 81 percent lower mean absolute error and 21 percent higher correlation than unimodal video baselines under challenging illumination and motion.
  • The same trained model works across multiple datasets without retraining the inference path.
  • Radar hardware is required only during the training phase, not at test time.
  • The adaptive gating prevents performance drops on low-quality teacher samples.

Where Pith is reading between the lines

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

  • The same frequency-domain alignment idea could be tested on other paired sensors that capture periodic body motion, such as wearable accelerometers paired with video.
  • If the policy network generalizes, it may reduce the need for perfectly synchronized training pairs in future cross-modal physiological work.
  • Deployment on phones or laptops becomes practical once the radar phase is complete, lowering hardware barriers for continuous monitoring.
  • Extending the validation split used for the meta-objective to include more diverse skin tones and ages would test whether the shared rhythm assumption holds universally.

Load-bearing premise

RGB and RF waveforms share a similar latent periodic rhythm in the frequency domain that can be distilled reliably without negative transfer when teacher quality varies across samples.

What would settle it

A controlled test in which distillation is forced on samples whose radar and video frequency spectra show clear mismatch in peak location or shape, then measuring whether video-only test error rises above the no-distillation baseline.

Figures

Figures reproduced from arXiv: 2606.28089 by Duoxun Tang, Jiangbo Yu, Jiyao Wang, Kaishun Wu, Qingyong Hu, Xiao Yang.

Figure 1
Figure 1. Figure 1: Motivation for our study. Left: rPPG measures pulsatile reflectance changes, whereas RF radar measures cardio-respiratory micro-displacements; their sensing physics yield mismatched time-domain morphology, making naive feature/time align￾ment prone to negative transfer. However, within the physiological band, both modal￾ities are modulated by the same periodic rhythm, motivating the distillation on fre￾que… view at source ↗
Figure 2
Figure 2. Figure 2: The overview of RPM-Distill. (A) In training, a fixed radar teacher provides guidance to a video student via distillation with three components. A spectral policy network predicts a per-sample gate and component weights from the student–teacher spectral relation map, and is optimized by bilevel meta-learning. (B) The process of bilevel meta optimization. A virtual student update is evaluated on a labeled v… view at source ↗
Figure 3
Figure 3. Figure 3: A visualization of component weights \protect \boldsymbol {\alpha } and distillation gate g distributional changes when training in different PhysDrive scenarios. We measure the difference by the paired t-test [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Visualization of pulse waveform examples generated by RPM-Distill and Fac￾torizePhys in light and dark skin tone groups of EquiPleth. sion/shift and spurious maxima, so emphasizing rhythm anchoring around the fundamental frequency is more effective than time-domain matching. Besides, [PITH_FULL_IMAGE:figures/full_fig_p014_4.png] view at source ↗
read the original abstract

Video-based remote physiological measurement (RPM) is highly accessible but remains fragile under varying illumination, skin tones, and motion. Radio frequency (RF) radar is largely invariant to illumination and appearance, providing complementary cardio-respiratory micro-motion cues; however, requiring radar at inference is often impractical due to its limited ubiquity and deployment overhead. We propose RPM-Distill, a physiology-guided cross-modal distillation framework that leverages synchronized radar only during training while retaining video-only inference. Our key observation is that although RGB and RF waveforms differ in sensing physics and time-domain morphology, they share similar latent periodic rhythm in the frequency domain. We thus distill physiology-structured spectral evidence to improve robustness, via losses that (i) anchor the fundamental peak, (ii) match the off-peak background distribution, and (iii) preserve spectral morphology and sharpness. To avoid negative transfer under sample-level teacher quality and alignment uncertainty, a spectral policy network predicts sample-level distillation gates and component weights from the student--teacher spectral relation map, learned with a meta bilevel objective on a small labeled validation split. Through extensive experiments in challenging conditions and cross-dataset settings, RPM-Distill brings 81\% MAE and 21\% correlation improvement over unimodal baselines. Code is at https://github.com/WJULYW/RPM-Distill.

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 RPM-Distill, a cross-modal distillation framework for remote physiological measurement (RPM) from video. RF radar data is used only at training time to distill physiology-structured spectral cues (fundamental peak anchoring, off-peak background matching, and morphology preservation) into a video model. A spectral policy network, trained via meta bilevel optimization on a small labeled validation split, predicts per-sample distillation gates and weights from the student-teacher spectral relation map to avoid negative transfer. The paper reports 81% MAE and 21% correlation gains over unimodal baselines under challenging conditions and cross-dataset evaluation, with code released.

Significance. If the empirical improvements and the generalization of the bilevel-learned policy hold under distribution shift, the work would offer a practical route to more robust video-only RPM by exploiting complementary RF sensing only during training. The frequency-domain distillation losses and adaptive policy address a recognized challenge in cross-modal physiological distillation. The public code release supports reproducibility.

major comments (2)
  1. [Abstract / §4] Abstract and §4 (method): the 81% MAE claim depends on the spectral policy network, trained via meta bilevel objective on a small validation split, correctly predicting per-sample gates and weights. No details are given on validation-split size, how the bilevel objective is optimized, or ablations isolating the policy's contribution versus the frequency losses; without these, it is impossible to verify that the reported cross-dataset gains are not artifacts of policy overfitting or negative transfer.
  2. [Abstract] Abstract: the core assumption that RGB and RF waveforms share reliably distillable latent periodic rhythm in the frequency domain despite differing sensing physics is stated but not accompanied by quantitative evidence (e.g., spectral similarity metrics or failure cases) that would demonstrate the assumption survives the cross-dataset shifts mentioned in the experiments.
minor comments (2)
  1. [Abstract] Abstract: dataset names, subject counts, illumination/motion conditions, and error-bar statistics are omitted, preventing assessment of the scale and statistical reliability of the claimed improvements.
  2. [Abstract] The abstract states 'extensive experiments' but provides no table or figure references; the manuscript should include explicit ablation tables on the three frequency-domain losses and on the policy network.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and commit to revisions that provide the requested details and evidence.

read point-by-point responses

  1. Referee: [Abstract / §4] Abstract and §4 (method): the 81% MAE claim depends on the spectral policy network, trained via meta bilevel objective on a small validation split, correctly predicting per-sample gates and weights. No details are given on validation-split size, how the bilevel objective is optimized, or ablations isolating the policy's contribution versus the frequency losses; without these, it is impossible to verify that the reported cross-dataset gains are not artifacts of policy overfitting or negative transfer.

    Authors: We agree that explicit details are needed to substantiate the policy network's role. In the revised manuscript we will report the validation split size used for the meta-objective, describe the bilevel optimization procedure (including the first-order approximation employed), and add ablations that compare the adaptive policy against fixed-weight and random-gating baselines. These additions will allow direct verification that the cross-dataset gains arise from the learned policy rather than overfitting or negative transfer. revision: yes

  2. Referee: [Abstract] Abstract: the core assumption that RGB and RF waveforms share reliably distillable latent periodic rhythm in the frequency domain despite differing sensing physics is stated but not accompanied by quantitative evidence (e.g., spectral similarity metrics or failure cases) that would demonstrate the assumption survives the cross-dataset shifts mentioned in the experiments.

    Authors: The consistent MAE and correlation gains under cross-dataset evaluation provide indirect support for the shared spectral structure. To supply direct quantitative evidence we will include, in the revision, average spectral similarity metrics (cosine similarity and Pearson correlation of normalized power spectral densities) computed on paired RGB-RF samples, together with an analysis of samples where the policy assigns low distillation weight due to misalignment. revision: yes

Circularity Check

0 steps flagged

No circularity: performance gains measured on external test distributions

full rationale

The paper introduces a cross-modal distillation method whose central components (frequency-domain losses and a meta-trained spectral policy network) are defined independently of the reported MAE/correlation metrics. The 81% MAE and 21% correlation figures are obtained on held-out test sets and cross-dataset evaluations, not by re-using fitted parameters or validation-split statistics as the evaluation target. No self-citations, uniqueness theorems, or ansatzes are invoked to close the derivation; the method is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Because only the abstract is available, the ledger is necessarily incomplete; the central claim rests on the unverified assumption that frequency-domain periodic rhythms are shared across modalities and that the policy network can be meta-learned without introducing new free parameters that dominate the result.

pith-pipeline@v0.9.1-grok · 5784 in / 1180 out tokens · 27693 ms · 2026-06-29T04:14:45.926368+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

74 extracted references · 4 linked inside Pith

  1. [1]

    npj Digital Medicine (2025)

    Acharya, B., Saakyan, W., Hammer, B., Drimalla, H.: The reliability of remote photoplethysmography under low illumination and elevated heart rates. npj Digital Medicine (2025)

    2025

  2. [2]

    In: Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems (CHI)

    Adib, F., Mao, H., Kabelac, Z., Katabi, D., Miller, R.C.: Smart homes that monitor breathing and heart rate. In: Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems (CHI). pp. 837–846 (2015)

    2015

  3. [3]

    IEEE Access 7, 54958–54968 (2019)

    Alizadeh, M., Shaker, G., De Almeida, J.C.M., Morita, P.P., Safavi-Naeini, S.: Remote monitoring of human vital signs using mm-wave fmcw radar. IEEE Access 7, 54958–54968 (2019)

    2019

  4. [4]

    IEEE Transactions on Pattern Analysis and Machine Intelligence 41(2), 423–443 (2018)

    Baltrušaitis, T., Ahuja, C., Morency, L.P.: Multimodal machine learning: A survey and taxonomy. IEEE Transactions on Pattern Analysis and Machine Intelligence 41(2), 423–443 (2018)

    2018

  5. [5]

    In: Proceed- ings of the IEEE/CVF international conference on computer vision

    Cho, J.H., Hariharan, B.: On the efficacy of knowledge distillation. In: Proceed- ings of the IEEE/CVF international conference on computer vision. pp. 4794–4802 (2019)

    2019

  6. [6]

    In: Proceedings of the AAAI Confer- ence on Artificial Intelligence

    Choi, J.H., Kang, K.B., Kim, K.T.: Fusion-vital: Video-rf fusion transformer for advanced remote physiological measurement. In: Proceedings of the AAAI Confer- ence on Artificial Intelligence. vol. 38, pp. 1344–1352 (2024)

    2024

  7. [7]

    IEEE transactions on biomedical engineering60(10), 2878–2886 (2013)

    De Haan, G., Jeanne, V.: Robust pulse rate from chrominance-based rppg. IEEE transactions on biomedical engineering60(10), 2878–2886 (2013)

    2013

  8. [8]

    npj Biosensing1(1), 5 (2024)

    Elgendi, M., Martinelli, I., Menon, C.: Optimal signal quality index for remote photoplethysmogram sensing. npj Biosensing1(1), 5 (2024)

    2024

  9. [9]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Feng, Q., Li, W., Lin, T., Chen, X.: Align-kd: Distilling cross-modal alignment knowledge for mobile vision-language large model enhancement. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 4178–4188 (2025)

    2025

  10. [10]

    In: Proceedings of the European Conference on Computer Vision (ECCV)

    Garcia, N.C., Morerio, P., Murino, V.: Modality distillation with multiple stream networks for action recognition. In: Proceedings of the European Conference on Computer Vision (ECCV). pp. 103–118 (2018)

    2018

  11. [11]

    In: Pro- ceedings of the 33rd ACM International Conference on Multimedia

    Ge, J., Sun, Z., Peng, W., Ying, C., Chen, Y., Ren, K., Li, X.: Evidential remote physiological measurement via uncertainty-aware fusion of video and rf. In: Pro- ceedings of the 33rd ACM International Conference on Multimedia. pp. 2467–2475 (2025)

    2025

  12. [12]

    Geng, Z., Guo, M.H., Chen, H., Li, X., Wei, K., Lin, Z.: Is attention better than matrix decomposition? In: International Conference on Learning Representations

  13. [13]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Gideon, J., Stent, S.: The way to my heart is through contrastive learning: Remote photoplethysmography from unlabelled video. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 3995–4004 (2021)

    2021

  14. [14]

    In: The Fourteenth International Conference on Learning Representations

    Guo, S., Zhao, H., Yang, S., Jiang, C., Cheng, Y., Peng, X., Shao, R., Tian, Z.: Multimodal dataset distillation via phased teacher models. In: The Fourteenth International Conference on Learning Representations

  15. [15]

    In:ProceedingsoftheIEEEconferenceoncomputervisionandpatternrecognition

    Gupta, S., Hoffman, J., Malik, J.: Cross modal distillation for supervision transfer. In:ProceedingsoftheIEEEconferenceoncomputervisionandpatternrecognition. pp. 2827–2836 (2016)

    2016

  16. [16]

    arXiv preprint arXiv:1503.02531 (2015)

    Hinton, G., Vinyals, O., Dean, J.: Distilling the knowledge in a neural network. arXiv preprint arXiv:1503.02531 (2015)

    Pith/arXiv arXiv 2015

  17. [17]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Hoffman, J., Gupta, S., Darrell, T.: Learning with side information through modal- ity hallucination. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 826–834 (2016) RPM-Distill 17

    2016

  18. [18]

    Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies8(4), 1–29 (2024)

    Hu, Q., Zhang, Q., Lu, H., Wu, S., Zhou, Y., Huang, Q., Chen, H., Chen, Y.C., Zhao, N.: Contactless arterial blood pressure waveform monitoring with mmwave radar. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies8(4), 1–29 (2024)

    2024

  19. [19]

    In: Proceedings of the 31st Annual International Conference on Mobile Computing and Networking

    Hu, Q., Zhou, Y., Wang, J., Huang, Z., Li, G., Xu, Q., Zhang, Q.: mmtremor: Prac- tical tremor monitoring for parkinson’s disease and essential tremor in daily life. In: Proceedings of the 31st Annual International Conference on Mobile Computing and Networking. pp. 498–512 (2025)

    2025

  20. [20]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Huo, F., Xu, W., Guo, J., Wang, H., Guo, S.: C2kd: Bridging the modality gap for cross-modal knowledge distillation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 16006–16015 (2024)

    2024

  21. [21]

    In: International conference on machine learning

    Ji, K., Yang, J., Liang, Y.: Bilevel optimization: Convergence analysis and en- hanced design. In: International conference on machine learning. pp. 4882–4892. PMLR (2021)

    2021

  22. [22]

    arXiv preprint arXiv:2411.01542 (2024)

    Joshi,J.,Agaian,S.S.,Cho,Y.:Factorizephys:Matrixfactorizationformultidimen- sional attention in remote physiological sensing. arXiv preprint arXiv:2411.01542 (2024)

    arXiv 2024

  23. [23]

    Electronics13(7), 1334 (2024)

    Joshi, J., Cho, Y.: Ibvp dataset: Rgb-thermal rppg dataset with high resolution signal quality labels. Electronics13(7), 1334 (2024)

    2024

  24. [24]

    Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Tech- nologies6(3), 1–30 (2022)

    Khan, U.M., Rigazio, L., Shahzad, M.: Contactless monitoring of ppg using radar. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Tech- nologies6(3), 1–30 (2022)

    2022

  25. [25]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Kim, S., Xiao, R., Georgescu, M.I., Alaniz, S., Akata, Z.: Cosmos: Cross-modality self-distillation for vision language pre-training. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 14690–14700 (2025)

    2025

  26. [26]

    IEEE Transactions on Microwave Theory and Techniques56(12), 3143– 3152 (2008)

    Li, C., Lin, J.: Random body movement cancellation in doppler radar vital sign detection. IEEE Transactions on Microwave Theory and Techniques56(12), 3143– 3152 (2008)

    2008

  27. [27]

    IEEE Transactions on Microwave Theory and Techniques61(5), 2046–2060 (2013)

    Li, C., Ling, J., Li, J., Lin, J.: A review on recent advances in doppler radar sensors for noncontact healthcare monitoring. IEEE Transactions on Microwave Theory and Techniques61(5), 2046–2060 (2013)

    2046

  28. [28]

    In: Proceedings of the 33rd ACM International Conference on Multimedia

    Li, H., Yang, P., Chen, J., Dong, L., Chen, Y., Wang, Q.: Mst-distill: Mixture of specialized teachers for cross-modal knowledge distillation. In: Proceedings of the 33rd ACM International Conference on Multimedia. pp. 1588–1597 (2025)

    2025

  29. [29]

    In: Proceedings of the AAAI conference on artificial intelligence

    Li, J., Yu, Z., Shi, J.: Learning motion-robust remote photoplethysmography through arbitrary resolution videos. In: Proceedings of the AAAI conference on artificial intelligence. vol. 37, pp. 1334–1342 (2023)

    2023

  30. [30]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Li, X., Chen, J., Zhao, G., Pietikainen, M.: Remote heart rate measurement from face videos under realistic situations. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 4264–4271 (2014)

    2014

  31. [31]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Liang, Q., Geng, R., Chen, J., Wang, H., Chen, Y., Hu, Y.: Spatial alignment and temporal matching adapter for video-radar remote physiological measurement. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 8623–8633 (2025)

    2025

  32. [32]

    arXiv preprint arXiv:1511.03643 (2015)

    Lopez-Paz, D., Bottou, L., Schölkopf, B., Vapnik, V.: Unifying distillation and privileged information. arXiv preprint arXiv:1511.03643 (2015)

    Pith/arXiv arXiv 2015

  33. [33]

    In: Proceed- ings of the IEEE/CVF International Conference on Computer Vision

    Martínez, J.C., Sukno, F.: Beatformer: Efficient motion-robust remote heart rate estimation through unsupervised spectral zoomed attention filters. In: Proceed- ings of the IEEE/CVF International Conference on Computer Vision. pp. 520–530 (2025) 18 J. Wang et al

    2025

  34. [34]

    IEEE transactions on biomedical circuits and systems12(3), 632–643 (2018)

    Mercuri, M., Liu, Y.H., Lorato, I., Torfs, T., Wieringa, F., Bourdoux, A., Van Hoof, C.: A direct phase-tracking doppler radar using wavelet independent component analysis for non-contact respiratory and heart rate monitoring. IEEE transactions on biomedical circuits and systems12(3), 632–643 (2018)

    2018

  35. [35]

    IEEE Transactions on Image Processing29, 2409–2423 (2019)

    Niu, X., Shan, S., Han, H., Chen, X.: Rhythmnet: End-to-end heart rate estima- tion from face via spatial-temporal representation. IEEE Transactions on Image Processing29, 2409–2423 (2019)

    2019

  36. [36]

    In: ProceedingsoftheIEEE/CVFConferenceonComputerVisionandPatternRecog- nition Workshops

    Nowara, E.M., McDuff, D., Veeraraghavan, A.: A meta-analysis of the impact of skin tone and gender on non-contact photoplethysmography measurements. In: ProceedingsoftheIEEE/CVFConferenceonComputerVisionandPatternRecog- nition Workshops. pp. 284–285 (2020)

    2020

  37. [37]

    In: Proceed- ings of the IEEE/CVF International Conference on Computer Vision

    Nowara, E.M., McDuff, D., Veeraraghavan, A.: The benefit of distraction: Denois- ing camera-based physiological measurements using inverse attention. In: Proceed- ings of the IEEE/CVF International Conference on Computer Vision. pp. 4955– 4964 (2021)

    2021

  38. [38]

    IEEE Transactions on Biomedi- cal Engineering58(1), 7–11 (2010)

    Poh, M.Z., McDuff, D.J., Picard, R.W.: Advancements in noncontact, multiparam- eter physiological measurements using a webcam. IEEE Transactions on Biomedi- cal Engineering58(1), 7–11 (2010)

    2010

  39. [39]

    Optics express 18(10), 10762–10774 (2010)

    Poh, M.Z., McDuff, D.J., Picard, R.W.: Non-contact, automated cardiac pulse measurements using video imaging and blind source separation. Optics express 18(10), 10762–10774 (2010)

    2010

  40. [40]

    IEEE Sensors Journal23(22), 28399–28410 (2023)

    Pokhrel, S.R., Kua, J., Satish, D., Williams, P., Zaslavsky, A., Loke, S.W., Choi, J.: Deakin rf-sensing: Experiments on correlated knowledge distillation for monitoring human postures with radios. IEEE Sensors Journal23(22), 28399–28410 (2023)

    2023

  41. [41]

    Romero, A., Ballas, N., Kahou, S.E., Chassang, A., Gatta, C., Bengio, Y.: Fitnets: Hints for thin deep nets (2015),https://arxiv.org/abs/1412.6550

    Pith/arXiv arXiv 2015

  42. [42]

    In: Proceedings of the 6th ACM International Conference on Mul- timedia in Asia

    Shan, L., Zhang, R., Chilukoti, S.V., Zhang, X., Lee, I., Hei, X.: Identitykd: Identity-wisecross-modalknowledgedistillationforpersonrecognitionviammwave radar sensors. In: Proceedings of the 6th ACM International Conference on Mul- timedia in Asia. pp. 1–7 (2024)

    2024

  43. [43]

    IEEE transactions on evolutionary computation22(2), 276–295 (2017)

    Sinha, A., Malo, P., Deb, K.: A review on bilevel optimization: From classical to evolutionary approaches and applications. IEEE transactions on evolutionary computation22(2), 276–295 (2017)

    2017

  44. [44]

    IEEE Journal of Biomedical and Health Informatics25(5), 1373–1384 (2021)

    Song, R., Chen, H., Cheng, J., Li, C., Liu, Y., Chen, X.: Pulsegan: Learning to generate realistic pulse waveforms in remote photoplethysmography. IEEE Journal of Biomedical and Health Informatics25(5), 1373–1384 (2021)

    2021

  45. [45]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Speth, J., Vance, N., Flynn, P., Czajka, A.: Non-contrastive unsupervised learning of physiological signals from video. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 14464–14474 (2023)

    2023

  46. [46]

    Špetlík, R., Franc, V., Matas, J.: Visual heart rate estimation with convolutional neuralnetwork.In:Proceedingsofthebritishmachinevisionconference,Newcastle, UK. pp. 3–6 (2018)

    2018

  47. [47]

    In: The 23rd IEEE International Symposium on Robot and Human Interactive Communication

    Stricker, R., Müller, S., Gross, H.M.: Non-contact video-based pulse rate measure- ment on a mobile service robot. In: The 23rd IEEE International Symposium on Robot and Human Interactive Communication. pp. 1056–1062. IEEE (2014)

    2014

  48. [48]

    In: Proceedings of the European conference on computer vision (ECCV)

    Sun, X., Xiao, B., Wei, F., Liang, S., Wei, Y.: Integral human pose regression. In: Proceedings of the European conference on computer vision (ECCV). pp. 529–545 (2018)

    2018

  49. [49]

    In: European Conference on Computer Vi- sion

    Sun,Z.,Li,X.:Contrast-phys:Unsupervisedvideo-basedremotephysiologicalmea- surement via spatiotemporal contrast. In: European Conference on Computer Vi- sion. pp. 492–510. Springer (2022) RPM-Distill 19

    2022

  50. [50]

    In: 2023 45th Annual International ConferenceoftheIEEEEngineeringinMedicine&BiologySociety(EMBC).pp.1–

    Tang, J., Chen, K., Wang, Y., Shi, Y., Patel, S., McDuff, D., Liu, X.: Mmpd: Multi-domain mobile video physiology dataset. In: 2023 45th Annual International ConferenceoftheIEEEEngineeringinMedicine&BiologySociety(EMBC).pp.1–

    2023

  51. [51]

    IEEE Transactions on Microwave Theory and Techniques64(6), 1937–1946 (2016)

    Tu, J., Hwang, T., Lin, J.: Respiration rate measurement under 1-d body mo- tion using single continuous-wave doppler radar vital sign detection system. IEEE Transactions on Microwave Theory and Techniques64(6), 1937–1946 (2016)

    1937

  52. [52]

    Neural Networks22(5-6), 544–557 (2009)

    Vapnik, V., Vashist, A.: A new learning paradigm: Learning using privileged infor- mation. Neural Networks22(5-6), 544–557 (2009)

    2009

  53. [53]

    Optics express16(26), 21434–21445 (2008)

    Verkruysse, W., Svaasand, L.O., Nelson, J.S.: Remote plethysmographic imaging using ambient light. Optics express16(26), 21434–21445 (2008)

    2008

  54. [54]

    ACM Transactions on Graphics (TOG)41(4), 1–14 (2022)

    Vilesov, A., Chari, P., Armouti, A., Harish, A.B., Kulkarni, K., Deoghare, A., Jalil- ian, L., Kadambi, A.: Blending camera and 77 ghz radar sensing for equitable, ro- bust plethysmography. ACM Transactions on Graphics (TOG)41(4), 1–14 (2022)

    2022

  55. [55]

    Accident Analysis & Prevention228, 108407 (2026)

    Wang, J., Li, W., Wang, Z., Ayas, S., Donmez, B., He, D., Wu, K.: Drowsydg-phys: Generalizable driver drowsiness estimation in conditional automated vehicles using physiological signals. Accident Analysis & Prevention228, 108407 (2026)

    2026

  56. [56]

    IEEE Transactions on Instrumentation and Measurement74, 1–14 (2024)

    Wang, J., Lu, H., Han, H., Chen, Y., He, D., Wu, K.: Generalizable remote phys- iological measurement via semantic-sheltered alignment and plausible style ran- domization. IEEE Transactions on Instrumentation and Measurement74, 1–14 (2024)

    2024

  57. [57]

    IEEE Journal of Biomedical and Health Informatics28(3), 1635–1643 (2023)

    Wang, J., Lu, H., Wang, A., Chen, Y., He, D.: Hierarchical style-aware domain generalization for remote physiological measurement. IEEE Journal of Biomedical and Health Informatics28(3), 1635–1643 (2023)

    2023

  58. [58]

    IEEE Transactions on Pattern Analysis and Machine Intelligence47(6), 4908–4925 (2025)

    Wang, J., Lu, H., Wang, A., Yang, X., Chen, Y., He, D., Wu, K.: Physmle: Generalizable and priors-inclusive multi-task remote physiological measurement. IEEE Transactions on Pattern Analysis and Machine Intelligence47(6), 4908–4925 (2025)

    2025

  59. [59]

    arXiv preprint arXiv:2605.01507 (2026)

    Wang, J., Wang, Y., Jiao, Y., Yang, X., He, D., Jafarnejad, S., Miranda-Moreno, L., Frank, R., Yu, J.: Mild: Mediator agent system with bidirectional percep- tion and multi-layered alignment for human-vehicle collaboration. arXiv preprint arXiv:2605.01507 (2026)

    Pith/arXiv arXiv 2026

  60. [60]

    IEEE Journal of Biomedical and Health Informatics28(12), 7090–7102 (2024)

    Wang, J., Wei, X., Lu, H., Chen, Y., He, D.: Condiff-rppg: Robust remote physio- logical measurement to heterogeneous occlusions. IEEE Journal of Biomedical and Health Informatics28(12), 7090–7102 (2024)

    2024

  61. [61]

    Advances in Neural Information Processing Systems38(2026)

    Wang, J., Yang, X., Hu, Q., Tang, J., Liu, C., He, D., Wang, Y., Chen, Y., Wu, K.: Physdrive: A multimodal remote physiological measurement dataset for in-vehicle driver monitoring. Advances in Neural Information Processing Systems38(2026)

    2026

  62. [62]

    International Journal of Computer Vision134(5), 199 (2026)

    Wang, J., Yang, X., Lu, H., He, D., Wu, K.: Align the gap: Prior-based unified multi-task remote physiological measurement framework for domain generalization and personalization. International Journal of Computer Vision134(5), 199 (2026)

    2026

  63. [63]

    IEEE Transactions on Biomedical Engineering67(5), 1462–1473 (2019)

    Wang, W., den Brinker, A.C., De Haan, G.: Discriminative signatures for remote- ppg. IEEE Transactions on Biomedical Engineering67(5), 1462–1473 (2019)

    2019

  64. [64]

    IEEE Transactions on Biomedical Engineering64(7), 1479–1491 (2016)

    Wang, W., Den Brinker, A.C., Stuijk, S., De Haan, G.: Algorithmic principles of remote ppg. IEEE Transactions on Biomedical Engineering64(7), 1479–1491 (2016)

    2016

  65. [65]

    In: 2020 15th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2020)

    Xi, L., Chen, W., Zhao, C., Wu, X., Wang, J.: Image enhancement for remote pho- toplethysmography in a low-light environment. In: 2020 15th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2020). pp. 1–7. IEEE (2020) 20 J. Wang et al

    2020

  66. [66]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Ying, C., Yang, H., Ge, J., Sun, Z., Cheng, X., Ren, K., Li, X.: Fusionphys: A flexible framework for fusing complementary sensing modalities in remote physio- logical measurement. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 9363–9373 (2025)

    2025

  67. [67]

    9th-12th September 2019, Cardiff, UK

    Yu, Z., Li, X., Zhao, G.: Remote photoplethysmograph signal measurement from facialvideosusingspatio-temporalnetworks.In:30thBritishMachineVisisonCon- ference: BMVC 2019. 9th-12th September 2019, Cardiff, UK. The British Machine Vision Conference (BMVC) (2019)

    2019

  68. [68]

    In: Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition

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

    2022

  69. [69]

    In: Interna- tional Conference on Learning Representations (2017)

    Zagoruyko, S., Komodakis, N.: Paying more attention to attention: Improving the performance of convolutional neural networks via attention transfer. In: Interna- tional Conference on Learning Representations (2017)

    2017

  70. [70]

    Zhang, K., Zhang, Z., Li, Z., Qiao, Y.: Joint face detection and alignment using multitaskcascadedconvolutionalnetworks.IEEESignalProcessingLetters23(10), 1499–1503 (2016)

    2016

  71. [71]

    In: Pro- ceedings of the IEEE conference on computer vision and pattern recognition

    Zhang, Y., Xiang, T., Hospedales, T.M., Lu, H.: Deep mutual learning. In: Pro- ceedings of the IEEE conference on computer vision and pattern recognition. pp. 4320–4328 (2018)

    2018

  72. [72]

    In: Proceedings of the 2018 Con- ference of the ACM Special Interest Group on Data Communication

    Zhao, M., Tian, Y., Zhao, H., Alsheikh, M.A., Li, T., Hristov, R., Kabelac, Z., Katabi, D., Torralba, A.: Rf-based 3d skeletons. In: Proceedings of the 2018 Con- ference of the ACM Special Interest Group on Data Communication. pp. 267–281 (2018)

    2018

  73. [73]

    In: Proceedings of the 19th ACM conference on embedded networked sensor systems

    Zheng, T., Chen, Z., Zhang, S., Cai, C., Luo, J.: More-fi: Motion-robust and fine- grained respiration monitoring via deep-learning uwb radar. In: Proceedings of the 19th ACM conference on embedded networked sensor systems. pp. 111–124 (2021)

    2021

  74. [74]

    In: Proceedings of the AAAI Conference on Artificial Intelligence

    Zou, B., Guo, Z., Hu, X., Ma, H.: Rhythmmamba: Fast, lightweight, and accurate remote physiological measurement. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 39, pp. 11077–11085 (2025)

    2025