Pith. sign in

REVIEW 3 major objections 8 minor 82 references

BeatFormer: Efficient motion-robust remote heart rate estimation through unsupervised spectral zoomed attention filters

T0 review · 3 major / 8 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read BeatFormer claims a 14.86k-parameter spectral attention model can match state-of-the-art motion-robust remote heart rate estimation, and can be trained without any PPG or heart-rate labels.

desk verdict Neat lightweight spectral rPPG architecture with honest ablations, but the SOTA claim is confounded by mismatched HR estimation and missing variance bars. read the letter →

arxiv 2507.14885 v1 pith:ECINAUCH submitted 2025-07-20 cs.CV

Joaquim Comas , Federico Sukno This is my paper · ORCID
classification cs.CV
keywords remotephotoplethysmographyheartrateestimationChirp-ZTransformspectralattentioncomplextransformercontrastivelearningmotionrobustnesslightweightneuralnetworks
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper introduces BeatFormer, a remote heart rate estimation model built from a spectral filter transformer and the Chirp-Z Transform. Its supervised version reaches cross-dataset mean absolute heart rate errors of 8.85 BPM (PURE to MMPD) and 8.98 BPM (UBFC-rPPG to MMPD), slightly below the prior best model RhythmFormer while using 14.86 thousand parameters and 181.73 million MACs. The paper further shows a version trained with no PPG or heart rate labels, using a new spectral contrastive learning objective, stays close to the supervised result (9.14 and 9.25 BPM). If these numbers hold, they indicate that most of the parameter cost in rPPG networks is unnecessary and that explicit physiological priors in the frequency domain can replace labels. The motivation is that handcrafted chrominance methods are robust to motion but linear, while deep models are accurate but heavy and label-hungry; BeatFormer is proposed as the middle path.

What carries the argument

The central machinery is the Chirp-Z Transform (CZT) as a zoomable spectral front end, the Zoomed Orthonormal Complex Attention (ZOCA) block, and the Spectral Contrastive Learning (SCL) objective. The CZT evaluates the z-transform along a spiral contour so it can concentrate the same number of bins on the 0.66-2.5 Hz heart rate band that the FFT would spread over the full 30 Hz range, giving roughly 13x finer resolution on short windows. ZOCA is a complex multi-head attention whose weight rows are constrained (first row fixed to the luminance-cancelling [1,1,1]/√3 direction, remaining rows orthonormal) so that the model acts as a learned spectral filter with physiological priors instead of a free transformer. SCL is a triplet hinge loss on the squared Earth Mover's Distance between power spectra of transformed versions of the same video (positive) and different videos (negative), using HSV/LAB conversions, temporal flipping, and random occlusion as transformations that are assumed to preserve pulsatile frequency content while altering motion noise.

What would settle it

Train BeatFormer-SCL on synthetic videos whose only periodic component is rigid head motion at a frequency inside the 0.66-2.5 Hz band, with no real cardiac signal; if the contrastive objective drives the network to output that motion frequency, the transformation-preservation assumption is false, whereas a flat or near-zero output would support it.

Watch

Extended reading notes

Core claim

BeatFormer is a lightweight transformer that operates directly on the frequency spectrum of spatially averaged facial skin RGB traces. It uses zoomed orthonormal complex attention (ZOCA), where the first row of each attention weight is fixed to the unit vector [1,1,1]/sqrt(3) and the remaining rows are constrained to be orthonormal, and an energy-measurement feed-forward that weights frequency subbands by their energy contribution relative to the input spectrum. The model is trained either supervised with a frequency-domain EMD loss plus orthonormality regularization, or without any labels using Spectral Contrastive Learning (SCL), which treats HSV/LAB color-space conversions, temporal flipping, and random occlusion of the same video as positive pairs and different videos as negatives. The paper's claim is that this combination reaches accuracy on the MMPD motion benchmark comparable to the best prior model while reducing the parameter count by over two orders of magnitude relative to RhythmFormer, and that SCL recovers most of the supervised gain without physiological labels.

Load-bearing premise

The load-bearing premise is that the four video transformations, HSV and LAB conversions, temporal flipping, and random occlusion, always keep the true heart-rate frequency intact while changing everything the model should ignore, so that forcing spectral similarity within a video teaches the cardiac signal rather than some other stable artifact.

Editorial extensions

If this is right

  • If BeatFormer's cross-dataset numbers reproduce, rPPG models under 15k parameters can match the accuracy of transformer models with hundreds of times more parameters, shifting the cost profile of on-device heart rate monitoring.
  • The SCL result implies that a pulse signal can be learned from unlabeled facial videos, so training sets for rPPG need not come with synchronized PPG contact sensors, removing a major data-collection bottleneck.
  • The CZT's zoomed spectral resolution means short temporal windows of around 300 frames can resolve heart rate frequencies that FFT-based pipelines smear, which matters for real-time applications with low latency.
  • The orthonormality prior suggests the classic chrominance intuition, cancelling intensity via [1,1,1]/√3 and separating color directions, can be embedded as a structural constraint in a deep network rather than as a fixed handcrafted projection.
  • The paper's motion-split results indicate the gap between handcrafted and deep methods on walking, rotation, and talking scenarios can be closed by spectral rather than spatiotemporal feature learning.

Reading between the lines

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

  • A direct extension of this result is that the same label-free spectral contrastive objective could apply to other periodic physiological signals, such as respiration rate or pulse transit time variability, wherever a transformation set can be designed that preserves the target periodicity.
  • Because BeatFormer uses spatially averaged RGB traces rather than per-pixel features, its robustness presumably comes from spectral filtering of the average trace; this suggests pixel-level motion compensation and spectral filtering are partially interchangeable sources of motion robustness, a trade-off the paper does not directly measure.
  • The paper's claim that phase information helps disentangle pulse from motion could be tested by ablating the imaginary-part components of the CZT spectrum; if phase is truly load-bearing, an FFT-based magnitude-only variant should degrade substantially more than the roughly 3 BPM gap reported in the ablation table.
  • One implicit consequence is that the transformation list in SCL is not fixed by the method; a practitioner facing a new population or camera type could add transformations, such as color-temperature shifts or compression artifacts, and keep the training label-free as long as the cardiac frequency remains invariant.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 8 minor

Summary. The paper proposes BeatFormer, a lightweight spectral transformer for remote photoplethysmography (rPPG) that operates on Chirp-Z Transform (CZT) spectra instead of raw temporal traces. The architecture combines a zoomed orthonormal complex attention (ZOCA) block with an energy-measurement feed-forward layer that filters RGB spectra. A second contribution, Spectral Contrastive Learning (SCL), uses hand-designed video transformations (HSV, LAB, temporal flipping, random occlusion) to train the model without PPG or heart-rate labels. The authors evaluate on PURE, UBFC-rPPG, and MMPD, reporting cross-dataset MAE of 8.85/8.98 BPM for the supervised model and 9.14/9.25 BPM for the label-free model, against 8.98/9.08 BPM for RhythmFormer, with about 14.86k parameters and 181.73 MACs. Ablations show that both CZT over FFT and ZOCA over plain complex attention improve MAE, and that the SCL loss approaches supervised performance.

Significance. If the reported results hold, the paper would demonstrate that a sub-15k-parameter spectral model can match state-of-the-art cross-dataset motion robustness and that a label-free training objective can come close to supervised performance. The design ideas are sensible: combining physiological priors from chrominance-based methods with a learned spectral filter is a plausible route to efficient and robust rPPG. The paper also benefits from ablations on the loss, the attention mechanism, the transform, and the SCL transformations, as well as from releasing code. The main limitations are that the headline comparison to prior work may not use the same heart-rate extraction pipeline, that no uncertainty quantification is provided, and that the label-free training assumption is not independently verified.

major comments (3)
  1. [§4.1, Table 1, Table 4] The central comparison may not be apples-to-apples. Section 4.1 states that BeatFormer's predicted rPPG is detrended, Butterworth-filtered, and converted to heart rate using CZT, but the comparator numbers in Table 1 are taken from prior publications that typically use FFT-based heart-rate extraction. This is load-bearing: Table 4 shows that replacing FFT with CZT changes MAE by roughly 4 to 5 BPM (14.50 to 9.15 without ZOCA; 13.07 to 8.85 with ZOCA), which is 30 to 40 times larger than the 0.13 BPM advantage claimed over RhythmFormer. The authors should rerun all baselines with the exact same evaluation pipeline, or at minimum report CZT-based MAE for the comparators and justify why their published numbers are directly comparable.
  2. [§4.2, Table 1] The reported differences are not supported by uncertainty quantification. Section 4.2 specifies a fixed random seed, 20 epochs, and batch size 2, and no confidence intervals or multi-seed means are reported. The performance gaps at stake are 0.13–0.29 BPM, which is small relative to typical seed-to-seed variation in deep rPPG training, especially with small training sets like PURE and UBFC-rPPG. The authors should provide results over multiple seeds with means and standard deviations, and ideally significance tests over videos, before claiming that BeatFormer-SL outperforms RhythmFormer or that SCL is comparable to SL.
  3. [§3.3, Figure 4] The SCL unsupervised claim rests on an unverified invariance assumption. The three assumptions in Section 3.3 state that the four video transformations preserve pulsatile frequency content while altering motion and other distortions, but this is asserted from physiological intuition rather than demonstrated. If the assumption fails in a new population, the label-free model has no corrective signal. The authors should provide empirical evidence, for example by showing that the predicted spectral peaks under each transformation agree with ground-truth heart rate on a labeled subset, or by measuring the consistency of the network's output across the four transformations. This is particularly important because Figure 4 shows that some transformations (e.g., spatial occlusion) behave differently across motion splits.
minor comments (8)
  1. [§3.1] The statement that CZT has 'approximately 13 times higher frequency resolution than FFT' at 30 fps appears inconsistent with the stated zoom band of 0.66–2.5 Hz: the ratio of full-band width to zoom-band width is 30/(2.5-0.66) ≈ 16.3, not 13. Please clarify the calculation.
  2. [§3.1, Eq. (3)] The matrix form of CZT is notationally inconsistent: A is described as an N×N diagonal matrix while the displayed diagonal uses entries A^{-0}, A^{-1}, ..., A^{-n}, and W is called an M×N Vandermonde matrix while the displayed exponent pattern is unclear. Please align the indices and dimensions.
  3. [§4.2] The phrase 'with a fixed random seed' does not identify the seed value or whether the reported numbers are from that single seed. Please report the seed used and, preferably, multiple seeds with summary statistics.
  4. [§4.2.2, Table 2] The computational-cost comparison says it follows the protocol of [82] and uses a FLOPs counting tool, but the table reports MACs. Please specify the input resolution and temporal length used for the count, and state whether the tool reports FLOPs or multiply-accumulate operations.
  5. [§4.3] The 'Impact of video transformations in SCL training' paragraph refers to Figure 3, but the relevant panel appears to be Figure 4; please correct the cross-reference.
  6. [§4.3] The sentence 'denoting a notable performance improvement (approximately 3 BPM)' understates the effect shown in Table 4: the improvement is about 4.2 BPM with ZOCA and about 5.4 BPM without ZOCA. Please use the actual values.
  7. [§3.2.2, Eq. (6)–(9)] The text says the complex matrix multiplication develops 'four complex attention blocks', but Eq. (9) contains eight sign-combined terms. Please clarify how the four blocks correspond to the eight terms.
  8. [§4.2.1] There is a typo in the text: 'RythmFormer' should be 'RhythmFormer' to match Table 1 and the reference list.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: BeatFormer's results are benchmarked against external MMPD ground truth; the only self-referential element is a minor, non-load-bearing citation of the authors' prior CZT work.

full rationale

The derivation chain is self-contained with respect to the claimed predictions. BeatFormer's filter weights are optimized either by a supervised loss against PPG ground truth or by the unsupervised spectral contrastive loss, and all headline metrics are computed against MMPD ground truth, which is external to the model and to the training datasets. The handcrafted components (CZT preprocessing, chrominance-inspired orthonormal attention constraints, energy-density filtering, and overlap-add reconstruction) are fixed priors taken from standard or published methods; they are inputs to the model rather than quantities derived from the target. The SCL invariance assumptions in Section 3.3 are empirical premises about video transformations, not a derivation that reduces to the predicted heart rate. The authors cite their own prior work [10, 11] for the CZT-based heart-rate estimator and for setting the CZT size, and the hyperparameters alpha and gamma in Eq. 17 are hand-tuned, but neither of these determines the main cross-dataset result by construction. Table 4's CZT-versus-FFT ablation shows that evaluation-pipeline choices can affect absolute MAE, but that is an ablation of the method rather than a circular reuse of the target. Accordingly, no specific equation or fitted parameter can be exhibited as a circular reduction. The separate question of whether comparator methods were evaluated with the same CZT-based HR extraction is an evaluation-protocol fairness issue, not a circularity in the paper's derivation chain.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim rests on five explicit physiological/domain assumptions and three hand-set parameters. None of these is a new physical entity. The assumptions are reasonable and inherited from prior rPPG literature, but they are not derived in the paper, so they set the boundary of validity for the claimed motion robustness and label-free training.

free parameters (3)
  • SCL margin gamma = 1
    Set based on preliminary experiments in Section 3.3, Eq. 16; affects the contrastive separation and thus the unsupervised result.
  • Orthonormal loss weight alpha = 0.5
    Empirically chosen in Section 3.3, Eq. 17 to balance SCL and ZOCA regularization.
  • Heart-rate zoom band = 0.66-2.5 Hz
    Chosen from rPPG literature as the physiological HR range in Section 3.1; if a subject's HR falls outside this band, the model cannot estimate it. This is a hand-set domain parameter, not fitted to the evaluation set.
assumptions (5)
  • domain assumption Pulse and motion signals have different relative amplitudes across RGB channels
    Used in the energy measurement S = |Z'|^2/|F|^2 in Section 3.2.3, Eq. 11, inherited from color-distortion filtering [68].
  • domain assumption The first attention row fixed to [1,1,1]/sqrt(3) and subsequent rows orthonormal encode the chrominance subspace that removes intensity variations
    ZOCA regularization in Section 3.2.2, Eq. 10, based on POS [67]; assumes this physiological prior transfers to unseen datasets.
  • domain assumption Pulse information is invariant to HSV and LAB color conversions, temporal flipping, and partial occlusion
    Foundation of SCL in Section 3.3, Figure 2; if transformations distort pulse, unsupervised learning has no label to detect it.
  • domain assumption The CZT zoom over 0.66-2.5 Hz captures all relevant heart rates at the given frame rates
    Section 3.1 restricts spectral analysis to the standard HR band, so out-of-band HR is undetectable.
  • domain assumption Spatially averaged skin-pixel RGB traces preserve enough pulsatile information
    Preprocessing in Section 3.2.1 follows standard rPPG practice but discards spatial detail, which may limit performance under large motions or occlusions.

how reviews work

0 comments
Cite this review

Pith. "Pith review of BeatFormer: Efficient motion-robust remote heart rate estimation through unsupervised spectral zoomed attention filters." pith.science (2026). https://pith.science/paper/ECINAUCH

@misc{pith2026250714885,
  author       = {Pith},
  title        = {Pith review of: BeatFormer: Efficient motion-robust remote heart rate estimation through unsupervised spectral zoomed attention filters},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ECINAUCH}},
  note         = {Machine review of arXiv:2507.14885}
}
read the original abstract

Remote photoplethysmography (rPPG) captures cardiac signals from facial videos and is gaining attention for its diverse applications. While deep learning has advanced rPPG estimation, it relies on large, diverse datasets for effective generalization. In contrast, handcrafted methods utilize physiological priors for better generalization in unseen scenarios like motion while maintaining computational efficiency. However, their linear assumptions limit performance in complex conditions, where deep learning provides superior pulsatile information extraction. This highlights the need for hybrid approaches that combine the strengths of both methods. To address this, we present BeatFormer, a lightweight spectral attention model for rPPG estimation, which integrates zoomed orthonormal complex attention and frequency-domain energy measurement, enabling a highly efficient model. Additionally, we introduce Spectral Contrastive Learning (SCL), which allows BeatFormer to be trained without any PPG or HR labels. We validate BeatFormer on the PURE, UBFC-rPPG, and MMPD datasets, demonstrating its robustness and performance, particularly in cross-dataset evaluations under motion scenarios.

Figures

Figures reproduced from arXiv: 2507.14885 by the authors.

Figure 1
Figure 1. BeatFormer overall structure. First, RGB traces are segmented with overlap and transformed into the frequency domain using [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. SCL Training Video Transformations (10-sec example): [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Performance comparison on MMPD motion scenarios. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: SCL video transformations impact in MMPD splits. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: MMPD cross-dataset inference example. Top: RGB [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

82 extracted references · 78 canonical work pages

  1. [1]

    Remote heart rate variability for emotional state monitoring

    Yannick Benezeth, Peixi Li, Richard Macwan, Keisuke Nakamura, Randy Gomez, and Fan Yang. Remote heart rate variability for emotional state monitoring. In EMBS Int. Conf. Biomed. Health Inform. BHI , pages 153–156. IEEE,

  2. [2]

    Alpine: Improving remote heart rate estima- tion using contrastive learning

    Lokendra Birla, Sneha Shukla, Anup Kumar Gupta, and Puneet Gupta. Alpine: Improving remote heart rate estima- tion using contrastive learning. In WACV, pages 5029–5038,

  3. [3]

    Unsupervised skin tissue seg- mentation for remote photoplethysmography.Pattern Recog- nit

    Serge Bobbia, Richard Macwan, Yannick Benezeth, Alamin Mansouri, and Julien Dubois. Unsupervised skin tissue seg- mentation for remote photoplethysmography.Pattern Recog- nit. Lett., 124:82–90, 2019. 6

    work page 2019

  4. [4]

    Orientation- conditioned facial texture mapping for video-based facial re- mote photoplethysmography estimation

    Sam Cantrill, David Ahmedt-Aristizabal, Lars Petersson, Hanna Suominen, and Mohammad Ali Armin. Orientation- conditioned facial texture mapping for video-based facial re- mote photoplethysmography estimation. In CVPR, pages 354–363, 2024. 2, 7

    work page 2024

  5. [5]

    Implicit neural models to extract heart rate from video

    Pradyumna Chari, Anirudh Bindiganavale Harish, Adnan Armouti, Alexander Vilesov, Sanjit Sarda, Laleh Jalilian, and Achuta Kadambi. Implicit neural models to extract heart rate from video. In ECCV, 2024. 2

    work page 2024

  6. [6]

    A simple framework for contrastive learn- ing of visual representations

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Ge- offrey Hinton. A simple framework for contrastive learn- ing of visual representations. In ICML, pages 1597–1607. PmLR, 2020. 2

    work page 2020

  7. [7]

    Deepphys: Video- based physiological measurement using convolutional atten- tion networks

    Weixuan Chen and Daniel McDuff. Deepphys: Video- based physiological measurement using convolutional atten- tion networks. In ECCV, pages 349–365, 2018. 2, 7, 8

    work page 2018

  8. [8]

    Effi- cient remote photoplethysmography with temporal derivative modules and time-shift invariant loss

    Joaquim Comas, Adria Ruiz, and Federico Sukno. Effi- cient remote photoplethysmography with temporal derivative modules and time-shift invariant loss. InCVPR, pages 2182– 2191, 2022. 2

    work page 2022

Show all 82 references
  1. [9]

    Physflow: Skin tone transfer for remote heart rate estimation through conditional normalizing flows

    Joaquim Comas, Antonia Alomar, Adria Ruiz, and Federico Sukno. Physflow: Skin tone transfer for remote heart rate estimation through conditional normalizing flows. BMVC,

  2. [10]

    Deep pulse-signal magnification for remote heart rate estimation in compressed videos

    Joaquim Comas, Adria Ruiz, and Federico Sukno. Deep pulse-signal magnification for remote heart rate estimation in compressed videos. arXiv preprint arXiv:2405.02652, 2024. 2, 6

    2024 arXiv

  3. [11]

    Deep adaptative spectral zoom for improved remote heart rate es- timation

    Joaquim Comas, Adria Ruiz, and Federico Sukno. Deep adaptative spectral zoom for improved remote heart rate es- timation. In FG, 2024. 1, 3

    2024

  4. [12]

    Prediction of vascular aging based on smart- phone acquired ppg signals

    Lorenzo Dall’Olio, Nico Curti, Daniel Remondini, Yosef Safi Harb, Folkert W Asselbergs, Gastone Castellani, and Hae-Won Uh. Prediction of vascular aging based on smart- phone acquired ppg signals. Scientific reports , 10:1–10,

  5. [13]

    Evaluation of biases in remote photoplethysmography methods

    Ananyananda Dasari, Sakthi Kumar Arul Prakash, L ´aszl´o A Jeni, and Conrad S Tucker. Evaluation of biases in remote photoplethysmography methods. NPJ digital medicine, 4(1): 91, 2021. 5

    2021

  6. [14]

    Robust pulse rate from chrominance-based rppg

    Gerard De Haan and Vincent Jeanne. Robust pulse rate from chrominance-based rppg. IEEE Trans. Biomed. Eng., 60(10): 2878–2886, 2013. 1, 2, 5, 7, 8

    2013

  7. [15]

    Dual-bridging with adversarial noise generation for domain adaptive rppg estimation

    Jingda Du, Si-Qi Liu, Bochao Zhang, and Pong C Yuen. Dual-bridging with adversarial noise generation for domain adaptive rppg estimation. In CVPR, pages 10355–10364,

  8. [16]

    The way to my heart is through contrastive learning: Remote photoplethysmogra- phy from unlabelled video

    John Gideon and Simon Stent. The way to my heart is through contrastive learning: Remote photoplethysmogra- phy from unlabelled video. In ICCV, pages 3995–4004,

  9. [17]

    Radiant: Better rppg estimation using signal embeddings and transformer

    Anup Kumar Gupta, Rupesh Kumar, Lokendra Birla, and Puneet Gupta. Radiant: Better rppg estimation using signal embeddings and transformer. In WACV, pages 4976–4986,

  10. [18]

    Squared earth movers distance loss for training deep neural networks on ordered-classes

    Le Hou, Chen-Ping Yu, and Dimitris Samaras. Squared earth movers distance loss for training deep neural networks on ordered-classes. In NIPS Workshop, 2017. 6

    2017

  11. [19]

    Augmentation of rppg benchmark datasets: Learning to re- move and embed rppg signals via double cycle consistent learning from unpaired facial videos

    Cheng-Ju Hsieh, Wei-Hao Chung, and Chiou-Ting Hsu. Augmentation of rppg benchmark datasets: Learning to re- move and embed rppg signals via double cycle consistent learning from unpaired facial videos. In ECCV, pages 372–

  12. [20]

    A neonatal dataset and benchmark for non-contact neonatal heart rate monitoring based on spatio-temporal neural net- works

    Bin Huang, Weihai Chen, Chun-Liang Lin, Chia-Feng Juang, Yuanping Xing, Yanting Wang, and Jianhua Wang. A neonatal dataset and benchmark for non-contact neonatal heart rate monitoring based on spatio-temporal neural net- works. Engineering Applications of Artificial Intelligen...

    2021

  13. [21]

    Introducing spectral attention for long-range dependency in time series forecasting

    Bong Gyun Kang, Dongjun Lee, HyunGi Kim, Dohyun Chung, and Sungroh Yoon. Introducing spectral attention for long-range dependency in time series forecasting. In NeurIPS. 2

  14. [22]

    Meta-rppg: Re- mote heart rate estimation using a transductive meta-learner

    Eugene Lee, Evan Chen, and Chen-Yi Lee. Meta-rppg: Re- mote heart rate estimation using a transductive meta-learner. In ECCV, pages 392–409. Springer, 2020. 2

    2020

  15. [23]

    Fnet: Mixing tokens with fourier transforms

    James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, and Santi- ago Ontanon. Fnet: Mixing tokens with fourier transforms. arXiv preprint arXiv:2105.03824, 2021. 2

    2021 arXiv

  16. [24]

    Learning motion- robust remote photoplethysmography through arbitrary res- olution videos

    Jianwei Li, Zitong Yu, and Jingang Shi. Learning motion- robust remote photoplethysmography through arbitrary res- olution videos. In AAAI, pages 1334–1342, 2023. 1, 2

    2023

  17. [25]

    Remote heart rate measurement from face videos under real- istic situations

    Xiaobai Li, Jie Chen, Guoying Zhao, and Matti Pietikainen. Remote heart rate measurement from face videos under real- istic situations. In CVPR, pages 4264–4271, 2014. 2, 6

    2014

  18. [26]

    Learning trajectory-aware transformer for video super- resolution

    Chengxu Liu, Huan Yang, Jianlong Fu, and Xueming Qian. Learning trajectory-aware transformer for video super- resolution. In CVPR, pages 5687–5696, 2022. 2

    2022

  19. [27]

    Spiking-physformer: Camera- based remote photoplethysmography with parallel spike- driven transformer

    Mingxuan Liu, Jiankai Tang, Yongli Chen, Haoxiang Li, Jiahao Qi, Siwei Li, Kegang Wang, Jie Gan, Yuntao Wang, and Hong Chen. Spiking-physformer: Camera- based remote photoplethysmography with parallel spike- driven transformer. Neural Networks, page 107128, 2025. 7

    2025

  20. [28]

    Fuseformer: Fusing fine-grained information in transformers for video inpainting

    Rui Liu, Hanming Deng, Yangyi Huang, Xiaoyu Shi, Lewei Lu, Wenxiu Sun, Xiaogang Wang, Jifeng Dai, and Hong- sheng Li. Fuseformer: Fusing fine-grained information in transformers for video inpainting. In ICCV, pages 14040– 14049, 2021. 2

    2021

  21. [29]

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

    Xin Liu, Josh Fromm, Shwetak Patel, and Daniel McDuff. Multi-task temporal shift attention networks for on-device contactless vitals measurement. NeurIPS, 33:19400–19411,

  22. [30]

    Metaphys: few-shot adaptation for non-contact physiological measurement

    Xin Liu, Ziheng Jiang, Josh Fromm, Xuhai Xu, Shwetak Patel, and Daniel McDuff. Metaphys: few-shot adaptation for non-contact physiological measurement. In CHIL, pages 154–163, 2021. 2

    2021

  23. [31]

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

    Xin Liu, Brian Hill, Ziheng Jiang, Shwetak Patel, and Daniel McDuff. Efficientphys: Enabling simple, fast and accurate camera-based cardiac measurement. In WACV, pages 5008– 5017, 2023. 2, 7

    2023

  24. [32]

    rppg-toolbox: Deep remote ppg toolbox

    Xin Liu, Girish Narayanswamy, Akshay Paruchuri, Xiaoyu Zhang, Jiankai Tang, Yuzhe Zhang, Roni Sengupta, Shwe- tak Patel, Yuntao Wang, and Daniel McDuff. rppg-toolbox: Deep remote ppg toolbox. NeurIPS, 36:68485–68510, 2023. 6

    2023

  25. [33]

    rppg-mae: Self-supervised pretraining with masked autoencoders for remote physiological mea- surements

    Xin Liu, Yuting Zhang, Zitong Yu, Hao Lu, Huanjing Yue, and Jingyu Yang. rppg-mae: Self-supervised pretraining with masked autoencoders for remote physiological mea- surements. IEEE Transactions on Multimedia, 2024. 2

    2024

  26. [34]

    Sgdr: Stochastic gradient descent with warm restarts

    Ilya Loshchilov and Frank Hutter. Sgdr: Stochastic gradient descent with warm restarts. In ICLR, 2022. 6

    2022

  27. [35]

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

    Hao Lu, Hu Han, and S Kevin Zhou. Dual-gan: Joint bvp and noise modeling for remote physiological measurement. In CVPR, pages 12404–12413, 2021. 2

    2021

  28. [36]

    Neuron structure modeling for generalizable remote physi- ological measurement

    Hao Lu, Zitong Yu, Xuesong Niu, and Ying-Cong Chen. Neuron structure modeling for generalizable remote physi- ological measurement. In CVPR, pages 18589–18599, 2023. 2

    2023

  29. [37]

    Medi- apipe: A framework for building perception pipelines

    Camillo Lugaresi, Jiuqiang Tang, Hadon Nash, Chris Mc- Clanahan, Esha Uboweja, Michael Hays, Fan Zhang, Chuo- Ling Chang, Ming Guang Yong, Juhyun Lee, et al. Medi- apipe: A framework for building perception pipelines. arXiv preprint arXiv:1906.08172, 2019. 6

    1906 arXiv

  30. [38]

    Physmamba: Effi- cient remote physiological measurement with slowfast tem- poral difference mamba

    Chaoqi Luo, Yiping Xie, and Zitong Yu. Physmamba: Effi- cient remote physiological measurement with slowfast tem- poral difference mamba. arXiv preprint arXiv:2409.12031,

    arXiv

  31. [39]

    Robustppg: camera-based robust heart rate estimation using motion cancellation

    Akash Kumar Maity, Jian Wang, Ashutosh Sabharwal, and Shree K Nayar. Robustppg: camera-based robust heart rate estimation using motion cancellation. Biomedical Optics Ex- press, 13(10):5447–5467, 2022. 1, 2

    2022

  32. [40]

    Rhythmnet: End-to-end heart rate estimation from face via spatial-temporal representation

    Xuesong Niu, Shiguang Shan, Hu Han, and Xilin Chen. Rhythmnet: End-to-end heart rate estimation from face via spatial-temporal representation. IEEE TIP, 29:2409–2423,

  33. [41]

    Near-infrared imaging photoplethysmogra- phy during driving

    Ewa M Nowara, Tim K Marks, Hassan Mansour, and Ashok Veeraraghavan. Near-infrared imaging photoplethysmogra- phy during driving. IEEE transactions on intelligent trans- portation systems, 23(4):3589–3600, 2020. 1

    2020

  34. [42]

    Motion matters: Neu- ral motion transfer for better camera physiological measure- ment

    Akshay Paruchuri, Xin Liu, Yulu Pan, Shwetak Patel, Daniel McDuff, and Soumyadip Sengupta. Motion matters: Neu- ral motion transfer for better camera physiological measure- ment. In WACV, pages 5933–5942, 2024. 1, 2

    2024

  35. [43]

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

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zem- ing Lin, Natalia Gimelshein, Luca Antiga, et al. Pytorch: An imperative style, high-performance deep learning library. NeurIPS, 32, 2019. 6

    2019

  36. [44]

    Hearttrack: Convolutional neural network for remote video-based heart rate monitoring

    Olga Perepelkina, Mikhail Artemyev, Marina Churikova, and Mikhail Grinenko. Hearttrack: Convolutional neural network for remote video-based heart rate monitoring. In CVPRW, pages 288–289, 2020. 2

    2020

  37. [45]

    Spa- tial temporal transformer network for skeleton-based action recognition

    Chiara Plizzari, Marco Cannici, and Matteo Matteucci. Spa- tial temporal transformer network for skeleton-based action recognition. In ICPR, pages 694–701. Springer, 2021. 2

    2021

  38. [46]

    Advancements in noncontact, multiparameter physiological measurements using a webcam

    Ming-Zher Poh, Daniel J McDuff, and Rosalind W Picard. Advancements in noncontact, multiparameter physiological measurements using a webcam. IEEE Trans. Biomed. Eng., 58(1):7–11, 2010. 2, 3

    2010

  39. [47]

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

    Ming-Zher Poh, Daniel J McDuff, and Rosalind W Picard. Non-contact, automated cardiac pulse measurements using video imaging and blind source separation. Optics express, 18(10):10762–10774, 2010. 2, 7

    2010

  40. [48]

    Spatiotempo- ral contrastive video representation learning

    Rui Qian, Tianjian Meng, Boqing Gong, Ming-Hsuan Yang, Huisheng Wang, Serge Belongie, and Yin Cui. Spatiotempo- ral contrastive video representation learning. InCVPR, pages 6964–6974, 2021. 2

    2021

  41. [49]

    Fcanet: Fre- quency channel attention networks

    Zequn Qin, Pengyi Zhang, Fei Wu, and Xi Li. Fcanet: Fre- quency channel attention networks. In ICCV, pages 783– 792, 2021. 2

    2021

  42. [50]

    The chirp z-transform algorithm

    L Rabiner, R W Schafer, and C Rader. The chirp z-transform algorithm. IEEE transactions on audio and electroacoustics, 17(2):86–92, 1969. 1, 3

    1969

  43. [51]

    Rs-rppg: robust self- supervised learning for rppg

    Marko Savic and Guoying Zhao. Rs-rppg: robust self- supervised learning for rppg. In 2024 IEEE 18th Interna- tional Conference on Automatic Face and Gesture Recogni- tion (FG), pages 1–10. IEEE, 2024. 2

    2024

  44. [52]

    Facenet: A unified embedding for face recognition and clus- tering

    Florian Schroff, Dmitry Kalenichenko, and James Philbin. Facenet: A unified embedding for face recognition and clus- tering. In CVPR, pages 815–823, 2015. 2

    2015

  45. [53]

    Pulsegan: Learning to generate realis- tic pulse waveforms in remote photoplethysmography

    Rencheng Song, Huan Chen, Juan Cheng, Chang Li, Yu Liu, and Xun Chen. Pulsegan: Learning to generate realis- tic pulse waveforms in remote photoplethysmography. IEEE J.Biomed.Health Inform., 25(5):1373–1384, 2021. 2

    2021

  46. [54]

    Visual heart rate estimation with convolutional neural network

    Radim ˇSpetl´ık, V ojtech Franc, and Jir´ı Matas. Visual heart rate estimation with convolutional neural network. InBMVC,

  47. [55]

    Non-contact video-based pulse rate measurement on a mo- bile service robot

    Ronny Stricker, Steffen M ¨uller, and Horst-Michael Gross. Non-contact video-based pulse rate measurement on a mo- bile service robot. In RO-MAN, pages 1056–1062. IEEE,

  48. [56]

    Contrast-phys: Unsuper- vised video-based remote physiological measurement via spatiotemporal contrast

    Zhaodong Sun and Xiaobai Li. Contrast-phys: Unsuper- vised video-based remote physiological measurement via spatiotemporal contrast. In ECCV, pages 492–510. Springer,

  49. [57]

    Contrast-phys+: Unsuper- vised and weakly-supervised video-based remote physiolog- ical measurement via spatiotemporal contrast

    Zhaodong Sun and Xiaobai Li. Contrast-phys+: Unsuper- vised and weakly-supervised video-based remote physiolog- ical measurement via spatiotemporal contrast. IEEE Trans- actions on Pattern Analysis and Machine Intelligence, 2024. 2

    2024

  50. [58]

    Heart rate measurement based on a time-lapse image

    Chihiro Takano and Yuji Ohta. Heart rate measurement based on a time-lapse image. Medical engineering & physics, 29(8):853–857, 2007. 2

    2007

  51. [59]

    Real- valued backpropagation is unsuitable for complex-valued neural networks

    Zhi-Hao Tan, Yi Xie, Yuan Jiang, and Zhi-Hua Zhou. Real- valued backpropagation is unsuitable for complex-valued neural networks. NeurIPS, 35:34052–34063, 2022. 4

    2022

  52. [60]

    Mmpd: Multi- domain mobile video physiology dataset

    Jiankai Tang, Kequan Chen, Yuntao Wang, Yuanchun Shi, Shwetak Patel, Daniel McDuff, and Xin Liu. Mmpd: Multi- domain mobile video physiology dataset. arXiv preprint arXiv:2302.03840, 2023. 1, 6

    2023 arXiv

  53. [61]

    An advanced detrending method with application to hrv analysis

    Mika P Tarvainen, Perttu O Ranta-Aho, and Pasi A Kar- jalainen. An advanced detrending method with application to hrv analysis. IEEE transactions on biomedical engineer- ing, 49(2):172–175, 2002. 6

    2002

  54. [62]

    Deep complex networks

    Chiheb Trabelsi, Olexa Bilaniuk, Ying Zhang, Dmitriy Serdyuk, Sandeep Subramanian, Joao Felipe Santos, Soroush Mehri, Negar Rostamzadeh, Yoshua Bengio, and Christopher J Pal. Deep complex networks. In ICLR, 2018. 2

    2018

  55. [63]

    On the benefits of alternative color spaces for noncontact heart rate measurements using standard red-green-blue cameras

    Gill R Tsouri and Zheng Li. On the benefits of alternative color spaces for noncontact heart rate measurements using standard red-green-blue cameras. Journal of biomedical op- tics, 20(4):048002–048002, 2015. 5

    2015

  56. [64]

    Attention is all you need

    A Vaswani. Attention is all you need. NeurIPS, 2017. 2

    2017

  57. [65]

    Re- mote plethysmographic imaging using ambient light

    Wim Verkruysse, Lars O Svaasand, and J Stuart Nelson. Re- mote plethysmographic imaging using ambient light. Optics express, 16(26):21434–21445, 2008. 2

    2008

  58. [66]

    Transphys: Transformer-based un- supervised contrastive learning for remote heart rate mea- surement

    Rui-Xuan Wang, Hong-Mei Sun, Rong-Rong Hao, Ang Pan, and Rui-Sheng Jia. Transphys: Transformer-based un- supervised contrastive learning for remote heart rate mea- surement. Biomedical Signal Processing and Control , 86: 105058, 2023. 2

    2023

  59. [67]

    Algorithmic principles of remote ppg

    Wenjin Wang, Albertus C den Brinker, Sander Stuijk, and Gerard De Haan. Algorithmic principles of remote ppg. IEEE Trans. Biomed. Eng. , 64(7):1479–1491, 2016. 1, 2, 3, 4, 7, 8

    2016

  60. [68]

    Color-distortion filtering for remote photo- plethysmography

    Wenjin Wang, Albertus C den Brinker, Sander Stuijk, and Gerard de Haan. Color-distortion filtering for remote photo- plethysmography. In FG, pages 71–78. IEEE, 2017. 1, 2, 4, 5

    2017

  61. [69]

    Robust heart rate from fitness videos.Phys- iological measurement, 38(6):1023, 2017

    Wenjin Wang, Albertus C den Brinker, Sander Stuijk, and Gerard de Haan. Robust heart rate from fitness videos.Phys- iological measurement, 38(6):1023, 2017. 1, 2

    2017

  62. [70]

    Complex gated recurrent neural networks

    Moritz Wolter and Angela Yao. Complex gated recurrent neural networks. NeurIPS, 31, 2018. 2

    2018

  63. [71]

    Motion robust remote photoplethysmography measure- ment during exercise for contactless physical activity inten- sity detection

    Yi-Chiao Wu, Li-Wen Chiu, Bing-Fei Wu, Linda Li-Chuan Lin, Tsai-Hsuan Ho, Meng-Liang Chung, and Shou-Fang Wu. Motion robust remote photoplethysmography measure- ment during exercise for contactless physical activity inten- sity detection. IEEE TIM, 72:1–14, 2023. 2

    2023

  64. [72]

    Complex transformer: A framework for modeling complex-valued sequence

    Muqiao Yang, Martin Q Ma, Dongyu Li, Yao-Hung Hu- bert Tsai, and Ruslan Salakhutdinov. Complex transformer: A framework for modeling complex-valued sequence. In ICASSP, pages 4232–4236. IEEE, 2020. 1, 2, 4, 8

    2020

  65. [73]

    Motion robust remote photo- plethysmography in cielab color space

    Yuting Yang, Chenbin Liu, Hui Yu, Dangdang Shao, Fran- cis Tsow, and Nongjian Tao. Motion robust remote photo- plethysmography in cielab color space. Journal of biomedi- cal optics, 21(11):117001–117001, 2016. 5

    2016

  66. [74]

    Yu, Xiao-Bai Li, and G

    Z. Yu, Xiao-Bai Li, and G. Zhao. Remote photoplethysmo- graph signal measurement from facial videos using spatio- temporal networks. In BMVC, 2019. 2, 7

    2019

  67. [75]

    Remote heart rate measurement from highly compressed facial videos: an end-to-end deep learning so- lution with video enhancement

    Zitong Yu, Wei Peng, Xiaobai Li, Xiaopeng Hong, and Guoying Zhao. Remote heart rate measurement from highly compressed facial videos: an end-to-end deep learning so- lution with video enhancement. In ICCV, pages 151–160,

  68. [76]

    Transrppg: Remote photoplethysmography transformer for 3d mask face presentation attack detection.IEEE Signal Pro- cessing Letters, 28:1290–1294, 2021

    Zitong Yu, Xiaobai Li, Pichao Wang, and Guoying Zhao. Transrppg: Remote photoplethysmography transformer for 3d mask face presentation attack detection.IEEE Signal Pro- cessing Letters, 28:1290–1294, 2021. 2

    2021

  69. [77]

    Physformer: facial video-based physiological measurement with temporal dif- ference transformer

    Zitong Yu, Yuming Shen, Jingang Shi, Hengshuang Zhao, Philip HS Torr, and Guoying Zhao. Physformer: facial video-based physiological measurement with temporal dif- ference transformer. In CVPR, pages 4186–4196, 2022. 2, 7

    2022

  70. [78]

    Physformer++: Facial video-based physiological measure- ment with slowfast temporal difference transformer

    Zitong Yu, Yuming Shen, Jingang Shi, Hengshuang Zhao, Yawen Cui, Jiehua Zhang, Philip Torr, and Guoying Zhao. Physformer++: Facial video-based physiological measure- ment with slowfast temporal difference transformer. IJCV, 131(6):1307–1330, 2023. 2

    2023

  71. [79]

    Demodulation based transformer for rppg generation and heart rate estimation

    Xiaobiao Zhang, Zhaoqiang Xia, Lili Liu, and Xiaoyi Feng. Demodulation based transformer for rppg generation and heart rate estimation. IEEE Signal Processing Letters, 2023. 2

    2023

  72. [80]

    Toward motion robust- ness: A masked attention regularization framework in re- mote photoplethysmography

    Pengfei Zhao, Qigong Sun, Xiaolin Tian, Yige Yang, Shuo Tao, Jie Cheng, and Jiantong Chen. Toward motion robust- ness: A masked attention regularization framework in re- mote photoplethysmography. In CVPR, pages 7829–7838,

  73. [81]

    Enhancing remote-ppg pulse extraction in disturbance scenarios utilizing spectral characteristics

    Kai Zhou, Simon Krause, Timon Blocher, and Wilhelm Stork. Enhancing remote-ppg pulse extraction in disturbance scenarios utilizing spectral characteristics. In CVPR, pages 280–281, 2020. 2

    2020

  74. [82]

    Rhythmformer: Extracting rppg signals based on hi- erarchical temporal periodic transformer

    Bochao Zou, Zizheng Guo, Jiansheng Chen, and Huimin Ma. Rhythmformer: Extracting rppg signals based on hi- erarchical temporal periodic transformer. arXiv preprint arXiv:2402.12788, 2024. 7

    2024 arXiv

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.