Pith. sign in

REVIEW 2 major objections 7 minor 53 references

You Only Flow Once: Calibrated and Real-Time Radar Pose Estimation with Multi-Hypothesis Normalizing Flows

T0 review · 2 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A conditional normalizing flow with a Laplace base distribution produces calibrated multi-hypothesis radar pose posteriors in a single forward pass, outperforming diffusion baselines in calibration on three benchmarks.

desk verdict Solid flow-based radar pose estimation with a real speed and calibration win over diffusion baselines; the headline 'calibrated posterior' claim is only supported per-coordinate, not jointly. read the letter →

arxiv 2608.09579 v1 pith:IKWJMIAY submitted 2026-08-10 cs.CV

Jonas Leo Mueller , Sebastian Hoefler , Dario Zanca , Naga Venkata Sai Jitin Jami , Thomas Altstidl , Bjoern M. Eskofier This is my paper · ORCID
classification cs.CV
keywords radarposeestimationmulti-hypothesisuncertaintynormalizingflowsmodelcalibrationmmWavepointcloudsRealNVPdiffusionbaselineshuman
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 argues that radar-based human pose estimation from sparse point clouds should output a calibrated distribution over plausible poses, not a single guess, because the sensor physics make the inverse problem many-to-one. It introduces MH-NFPG, a conditional normalizing flow that turns a Laplace base distribution into a flexible pose posterior in one forward pass, using a spatiotemporal transformer backbone and a learned multivariate Gaussian prior as conditioning. Across MM-Fi, mmRadPose, and mRI, the paper reports that this flow consistently matches nominal coverage levels where diffusion baselines undercover, with expected calibration error reduced by 34% to 85% and inference about 20 times faster. The claimed payoff is that a safety-critical system can trust its uncertainty estimates in real time.

What carries the argument

The load-bearing object is the conditional Real NVP normalizing flow. It is a stack of affine coupling layers that bijectively map poses to latent variables; each layer's scale and translation come from a conditioning network fed by transformer backbone features and past prior predictions. The log-likelihood is exact via the change-of-variables formula, so training minimizes negative log-likelihood directly, and inverse sampling produces all hypotheses in parallel. The conditioning prior, a heteroscedastic multivariate Gaussian, supplies per-input covariance that captures anisotropic, input-dependent radar uncertainty and is itself a baseline for calibration.

What would settle it

Re-run the evaluation on mmRadPose using optical motion-capture ground truth and compute joint credible regions over the full keypoint pose, for instance with a multivariate depth-based probability integral transform, instead of per-coordinate intervals. If empirical joint coverage falls well below nominal 95% while per-coordinate expected calibration error stays low, or if coverage degrades sharply when the hypothesis count is reduced from 200, then the reported marginal calibration would not support the claim of calibrated 3D pose distributions.

Watch

Extended reading notes

Core claim

The paper claims that MH-NFPG produces empirically calibrated multi-hypothesis pose distributions from radar point clouds in a single forward pass, outperforming diffusion-based alternatives in calibration on all three benchmarks and in pose accuracy on two. The mechanism is a two-phase pipeline: Phase 1 learns a heteroscedastic multivariate Gaussian prior over poses from a permutation-invariant spatiotemporal transformer, and Phase 2 freezes it and trains a conditional Real NVP normalizing flow that maps an independent Laplace base, each component with scale $1/\sqrt{2}$, into the posterior, conditioned on backbone features and temporal pose predictions. Because the flow is invertible, training uses the exact log-likelihood through the change-of-variables formula, and because inference is a single parallel inverse pass, all hypotheses are generated at once. The paper reports 80 frames per second inference, 95% coverage close to nominal on all benchmarks, and robustness to held-out environments, concluding that normalizing flows are a practical real-time alternative to diffusion models for uncertainty-aware radar pose estimation.

Load-bearing premise

The claim that the posterior is calibrated rests on per-coordinate marginal calibration, measured by the probability-integral-transform based expected calibration error; the full 3D pose's joint calibration and correlation structure are never evaluated, so if those matter for safety-critical use, the central claim would need additional evidence.

Editorial extensions

If this is right

  • Real-time uncertainty-aware radar pose estimation becomes feasible: 200 hypotheses can be drawn at 80 frames per second, which suits safety-relevant deployment such as human-robot collaboration and rehabilitation monitoring.
  • Sequential denoising is no longer required for multi-hypothesis radar pose models, so the main computational advantage of diffusion over flows is removed.
  • Calibration persists under distribution shift: on a held-out environment the reported expected calibration error stays low, which would support cross-environment deployment without post-hoc recalibration.
  • Calibrated uncertainty does not cost accuracy: the paper reports that the flow matches or improves on the strongest baseline for MPJPE and PA-MPJPE across the three datasets.
  • Exact-likelihood training gives calibration without a separate recalibration set, simplifying the training pipeline relative to diffusion baselines.

Reading between the lines

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

  • A testable extension is to apply the same conditioning architecture to other ill-posed inverse problems with sparse measurements, such as radar-based body shape reconstruction or single-image 3D lifting, where multi-hypothesis calibration is equally critical.
  • The reported per-joint sharpness pattern, with hands, feet, and head showing higher uncertainty than larger reflective body parts, suggests the model has learned physically meaningful structure; a follow-up could exploit that structure to sharpen intervals without losing coverage.
  • Because calibration is reported per coordinate, the model's uncertainty about whole-pose consistency is unexamined; a multivariate calibration check would determine whether the correlation structure of the posterior can be trusted for full-skeleton safety decisions.
  • The speed and calibration claims depend on sampling 200 hypotheses; deployment studies should verify how quickly coverage degrades with fewer samples before choosing a smaller budget.
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

2 major / 7 minor

Summary. The paper proposes MH-NFPG, a two-stage pipeline for radar point-cloud human pose estimation. Stage 1 trains a spatiotemporal transformer to parameterize a heteroscedastic multivariate Gaussian prior over 3D poses; Stage 2 freezes this prior and trains a conditional Real NVP normalizing flow with a Laplace base distribution, conditioned on backbone features and temporal predictions from the prior. At inference, 200 pose hypotheses are produced in parallel by a single inverse flow pass. Experiments on MM-Fi, mmRadPose, and mRI compare against deterministic PointTransformer, stochastic mmDiff, and DiffPose adapted to radar, reporting MPJPE, PA-MPJPE, marginal PIT-based ECE, interval coverage, and sharpness. The paper claims calibrated posteriors, 14-21x faster inference than diffusion baselines, and accuracy parity or improvement.

Significance. The technical construction is sound and the evaluation is unusually thorough for this area: three public benchmarks, standard plus leave-one-environment-out splits, an ablation covering coupling depth, conditioning architecture, base distribution, NF family, and temporal context, and a component-wise runtime breakdown. The exact-likelihood training of the flow and the parallel sampling strategy are appropriate, and the cross-environment calibration results, if reproducible, would be a useful step toward uncertainty-aware radar pose estimation. The main gap is that all calibration evidence is marginal (per-coordinate), whereas the central claim is about calibrated full-pose posteriors; this needs to be closed before the headline claim is supported.

major comments (2)
  1. [Section 3.3, Eq. (11); Tables 1-4] The calibration evidence is entirely marginal. ECE is computed from per-coordinate PIT values u_i = Fhat(y*_i), and the table header defines @X% as the proportion of ground-truth joints inside the per-joint credible interval. A model can be perfectly calibrated in every coordinate while badly miscalibrated for the full K-keypoint pose; for K=17, independent marginals give roughly 0.95^17 ≈ 0.42 full-pose coverage for 95% per-joint intervals. Since Section 4.1 states 'MH-NFPG consistently produces well-calibrated posteriors' and Section 5 concludes it 'delivers the calibrated uncertainty that a safety-critical system needs,' the paper must either add a joint calibration evaluation (e.g., a multivariate PIT based on the Rosenblatt transform, or coverage of full-pose credible regions computed from the flow samples) or explicitly restrict the claim to marginal calibration. This is the load-bearing gap between the reported metric and the central claim.
  2. [Tables 1-6] All accuracy and calibration numbers come from a single run with no repeated-seed variation or confidence intervals. The headline ECE gaps are large, but the mRI MPJPE comparison (8.269 vs 8.225 cm) is within the range one would expect from run-to-run variation, and PIT-based ECE from 200 hypotheses has sampling noise. Please report mean and standard deviation over at least three seeds, or bootstrap confidence intervals, for the central comparisons in Tables 1-5.
minor comments (7)
  1. [Abstract] The abstract claims 'over 20x faster inference', but Table 5 shows MG-Prior + DiffPose is 173.2/12.46 ≈ 13.9x slower than MH-NFPG; only mmDiff (η=1) is more than 20x slower. The contribution bullet already states 14-21x; align the abstract with the actual range.
  2. [Section 3.3, Eq. (11)] The text says C=100 levels p_j ∈ [0.01, 0.99]; with a 0.01 step this is 99 levels. Define the grid explicitly.
  3. [Section 3.3] The symbol N is used both for the number of test samples in Eq. (11) and for the number of hypotheses (200) elsewhere; rename one to avoid ambiguity.
  4. [Figures 1 and 3 captions] The captions refer to 'confidence ellipsoids'; these are per-joint marginal ellipsoids, not joint credible regions for the full pose. The captions should say so.
  5. [References] References [25] and [26] are duplicates of the same paper ('Human pose regression with residual log-likelihood estimation'); one should be removed.
  6. [Table 6] The default L=8 configuration appears only in the 'Ours' column, so the effect of L=8 is not visible in the coupling-layer columns; add an L=8 column for readability.
  7. [Section 3.2] Clarify whether the mRI glitch filtering (1% of training frames, 0.9% of test frames) is applied identically to all baselines.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: MH-NFPG's calibration claim is an empirical result of exact-likelihood training on held-out ground truth, with only minor non-load-bearing self-citation and a marginal-vs-joint metric gap.

full rationale

MH-NFPG is trained by minimizing the exact negative log-likelihood of ground-truth poses (Eq. 8) and evaluated on held-out test splits against external baselines (mmDiff, DiffPose, PointTransformer). The reported calibration metrics (PIT ECE in Eq. 11, coverage in Tables 1-4) compare the model's predictive samples with ground truth on test data, so the headline calibration result is measured, not constructed. The multivariate Gaussian prior is trained in Phase 1 on the same labels, but it is used as an internal conditioning signal for the flow; the flow itself is not defined as the prior, and the paper explicitly compares MG-Prior + NFPG with the standalone MG-Prior, so no fitted input is renamed as a prediction. The Laplace base distribution and Real NVP coupling layers are design choices that are ablated (Table 6), not assumptions that entail the target result. Self-citations [31] and [32] provide background and a baseline (RadProPoser) but no load-bearing argument reduces to them. The only noteworthy gap is that the 'calibrated posterior' claim is operationalized per coordinate (per-joint PIT and per-joint coverage), whereas the conclusion speaks of calibrated full-pose distributions; this is a metric-validity concern, not a circular reduction, and does not affect the independence of the empirical comparison.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard ML assumptions and on the specific choice of measuring calibration marginally per coordinate. No new physical entities or forces are introduced. The main free parameters are two loss weights tuned on a validation split and several architecture choices selected by ablation.

free parameters (5)
  • gamma (variance regularizer weight) = 1
    Selected by grid search on the MM-Fi validation split (Appendix B) to balance accuracy and calibration.
  • lambda_KL (KL divergence weight) = 15
    Selected by grid search on the MM-Fi validation split (Appendix B).
  • number of coupling layers L = 8
    Chosen by ablation as the default; deeper flows slightly improve MPJPE but degrade calibration.
  • temporal context frames n_t = 6
    Chosen for conditioning the flow; ablation shows temporal conditioning helps accuracy.
  • number of hypotheses N = 200
    Used for all probabilistic models, matching prior work; Figure 4 shows diminishing returns beyond approximately 100.
assumptions (4)
  • domain assumption Marginal per-coordinate calibration (PIT-based ECE) is an adequate measure of posterior calibration.
    The paper defines calibration via per-coordinate empirical CDFs in Section 3.3 and never evaluates joint calibration of the full pose.
  • domain assumption The NF posterior, trained with exact NLL, is expressive enough to represent the true ambiguity after affine coupling of a Laplace base.
    The paper relies on this in Section 3.1.1; ablations test some variants but not all.
  • domain assumption The conditioning features from the MG prior capture the relevant aleatoric uncertainty and temporal context.
    The NF is conditioned on these features (Section 3.1.1); if they are insufficient, the posterior would be miscalibrated.
  • domain assumption Ground-truth poses in the datasets are sufficiently accurate for calibration evaluation, modulo the disclosed glitch filtering.
    The paper filters 1% of mRI training and 0.9% of test frames with large pose jumps (Section 3.2).

how reviews work

0 comments
Cite this review

Pith. "Pith review of You Only Flow Once: Calibrated and Real-Time Radar Pose Estimation with Multi-Hypothesis Normalizing Flows." pith.science (2026). https://pith.science/paper/IKWJMIAY

@misc{pith2026260809579,
  author       = {Pith},
  title        = {Pith review of: You Only Flow Once: Calibrated and Real-Time Radar Pose Estimation with Multi-Hypothesis Normalizing Flows},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IKWJMIAY}},
  note         = {Machine review of arXiv:2608.09579}
}
read the original abstract

Sparse and noisy millimeter-wave radar point cloud observations often correspond to multiple plausible human poses, making deterministic pose estimation fundamentally ill-posed. Yet existing radar methods remain deterministic, collapsing this ambiguity into a single estimate. Diffusion-based alternatives can model multi-hypothesis distributions but require costly sequential denoising for each distribution sample and lack calibrated uncertainty. We propose Multi-Hypothesis Normalizing Flow Pose Generator (MH-NFPG), which models pose distributions from radar point clouds using a conditional normalizing flow. Specifically, we combine a spatiotemporal transformer backbone with a normalizing flow that transforms a Laplace base distribution into an expressive posterior, generated in parallel through a single forward pass. Leveraging this efficiency, we outperform diffusion-based alternatives in calibration across three radar benchmarks (MM-Fi, mmRadPose, mRI), improve pose accuracy on two, and match it on the third, while achieving over 20x faster inference for applications and reducing calibration error by up to 85%. We find that calibration degrades substantially for diffusion models, whereas our flow-based approach maintains reliable coverage, also in cross-environment settings. These results demonstrate normalizing flows as a practical alternative to diffusion models for real-time, uncertainty-aware radar pose estimation. Our code will be made publicly available.

Figures

Figures reproduced from arXiv: 2608.09579 by the authors.

Figure 1
Figure 1. MH-NFPG produces calibrated 3D pose distributions from radar point clouds in a single forward pass, with predictions (colored), ground truth (dashed), and 50% confidence ellipsoids on mmRadPose (blue), mRI (green), and MM-Fi (purple). 23, 44, 49] whose safety-critical use makes confident errors costly. An overconfident pose can corrupt a clinician’s as￾sessment in rehabilitation monitoring [32], or drive a robot int… view at source ↗
Figure 2
Figure 2. Overview of MH-NFPG. Phase 1 trains a transformer encoder on radar point cloud sequences to parameterize a heteroscedastic [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Qualitative comparison on mmRadPose. Predicted [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Effect of the number of hypotheses on MPJPE and cali [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Qualitative results of MH-NFPG across mmRadPose, mRI, and MM-Fi. Predicted poses with per-joint 90% confidence ellipsoids [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: Extended qualitative comparison on mmRadPose. Predictions (black) and ground truth (red) with 90% confidence ellipsoids for [PITH_FULL_IMAGE:figures/full_fig_p016_6.png]
Figure 7
Figure 7. Figure 7: Reliability diagram on mmRadPose. MH-NFPG (blue) [PITH_FULL_IMAGE:figures/full_fig_p017_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 45 canonical work pages

  1. [1]

    A review of uncertainty quantification in deep learning: Techniques, applications and challenges.Informa- tion fusion, 76:243–297, 2021

    Moloud Abdar, Farhad Pourpanah, Sadiq Hussain, Dana Rezazadegan, Li Liu, Mohammad Ghavamzadeh, Paul Fieguth, Xiaochun Cao, Abbas Khosravi, U Rajendra Acharya, et al. A review of uncertainty quantification in deep learning: Techniques, applications and challenges.Informa- tion fusion, 76:243–297, 2021

    work page 2021

  2. [2]

    Sizhe An and Umit Y . Ogras. Fast and scalable human pose estimation using mmWave point cloud. InProceedings of the 59th ACM/IEEE Design Automation Conference, pages 889–894, 2022

    work page 2022

  3. [3]

    mRI: Multi-modal 3D human pose estimation dataset using mmWave, RGB-D, and inertial sensors.Advances in Neural Information Processing Systems (NeurIPS), 2022

    Sizhe An, Yin Li, and Umit Ogras. mRI: Multi-modal 3D human pose estimation dataset using mmWave, RGB-D, and inertial sensors.Advances in Neural Information Processing Systems (NeurIPS), 2022

    work page 2022

  4. [4]

    Analyzing in- verse problems with invertible neural networks

    L Ardizzone, J Kruse, C Rother, and U K ¨othe. Analyzing in- verse problems with invertible neural networks. 7th int. In Conf. on Learning Representations, ICLR 2019 (New Or- leans, LA), 2019

    work page 2019

  5. [5]

    The need for uncertainty quantification in machine- assisted medical decision making.Nature Machine Intelli- gence, 1(1):20–23, 2019

    Edmon Begoli, Tanmoy Bhattacharya, and Dimitri Kusne- zov. The need for uncertainty quantification in machine- assisted medical decision making.Nature Machine Intelli- gence, 1(1):20–23, 2019

    work page 2019

  6. [6]

    Plausible uncertainties for human pose regression

    Lennart Bramlage, Michelle Karg, and Crist ´obal Curio. Plausible uncertainties for human pose regression. InPro- ceedings of the IEEE/CVF International Conference on Computer Vision, pages 15133–15142, 2023

    work page 2023

  7. [7]

    mmBody benchmark: 3D body re- construction dataset and analysis for millimeter wave radar

    Anjun Chen, Xiangyu Wang, Shaohao Zhu, Yanxu Li, Jim- ing Chen, and Qi Ye. mmBody benchmark: 3D body re- construction dataset and analysis for millimeter wave radar. ACM Multimedia, 2022

    work page 2022

  8. [8]

    Enhancing skeletal pose estimation from mmWave point clouds through uncertainty reduction

    Hsin-Che Chiang, Guan-Hua Li, Fan Wang, Shervin Shir- mohammadi, and Cheng-Hsin Hsu. Enhancing skeletal pose estimation from mmWave point clouds through uncertainty reduction. InProceedings of the 5th International Workshop on Human-centric Multimedia Analysis, pages 45–53, 2024

    work page 2024

Show all 53 references
  1. [9]

    Density estimation using Real NVP

    Laurent Dinh, Jascha Sohl-Dickstein, and Samy Bengio. Density estimation using Real NVP. InInternational Con- ference on Learning Representations (ICLR), 2017

    2017

  2. [10]

    Neural spline flows.Advances in Neural Infor- mation Processing Systems, 32, 2019

    Conor Durkan, Artur Bekasov, Iain Murray, and George Pa- pamakarios. Neural spline flows.Advances in Neural Infor- mation Processing Systems, 32, 2019

    2019

  3. [11]

    Eskofier, and Martin V ossiek

    Lukas Engel, Jonas Mueller, Eduardo Javier Feria Ren- don, Eva Dorschky, Daniel Krauss, Ingrid Ullmann, Bjo- ern M. Eskofier, and Martin V ossiek. Advanced millimeter wave radar-based human pose estimation enabled by a deep learning neural network trained with optical motion c...

    2025

  4. [12]

    Dif- fusion model is a good pose estimator from 3D RF-vision

    Junqiao Fan, Jianfei Yang, Yuecong Xu, and Lihua Xie. Dif- fusion model is a good pose estimator from 3D RF-vision. InEuropean Conference on Computer Vision (ECCV), pages 1–18. Springer, 2024

    2024

  5. [13]

    DiffPose: Spatiotemporal diffusion model for video-based human pose estimation

    Runyang Feng, Yixing Gao, Tze Ho Elden Tse, Xueqing Ma, and Hyung Jin Chang. DiffPose: Spatiotemporal diffusion model for video-based human pose estimation. InIEEE/CVF International Conference on Computer Vision (ICCV), pages 14861–14872, 2023

    2023

  6. [14]

    DiffPose: Toward more reliable 3D pose estimation

    Jia Gong, Lin Geng Foo, Zhipeng Fan, Qiuhong Ke, Hossein Rahmani, and Jun Liu. DiffPose: Toward more reliable 3D pose estimation. InIEEE/CVF Conference on Computer Vi- sion and Pattern Recognition (CVPR), pages 13041–13051, 2023

    2023

  7. [15]

    Gundavarapu, Divyansh Srivastava, Rahul Mitra, Abhishek Sharma, and Arjun Jain

    Nitesh B. Gundavarapu, Divyansh Srivastava, Rahul Mitra, Abhishek Sharma, and Arjun Jain. Structured aleatoric un- certainty in human pose estimation. InCVPR Workshops, 2019

    2019

  8. [16]

    Propose: Probabilistic 3d human pose estimation with instance-level distribution and normalizing flow

    Ju-Min Han, Jun-Hee Kim, and Seong-Whan Lee. Propose: Probabilistic 3d human pose estimation with instance-level distribution and normalizing flow. InAAAI Conference on Artificial Intelligence, 2025

    2025

  9. [17]

    Normalizing flows for human pose anomaly detection

    Or Hirschorn and Shai Avidan. Normalizing flows for human pose anomaly detection. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 13545– 13554, 2023

    2023

  10. [18]

    Rt-pose: A 4d radar tensor-based 3d human pose estimation and localization benchmark

    Yuan-Hao Ho, Jen-Hao Cheng, Sheng Yao Kuan, Zhongyu Jiang, Wenhao Chai, Hsiang-Wei Huang, Chih-Lung Lin, and Jenq-Neng Hwang. Rt-pose: A 4d radar tensor-based 3d human pose estimation and localization benchmark. In European Conference on Computer Vision, pages 107–125. Springer, 2024

    2024

  11. [19]

    DiffPose: Multi- hypothesis human pose estimation using diffusion models

    Karl Holmquist and Bastian Wandt. DiffPose: Multi- hypothesis human pose estimation using diffusion models. InIEEE/CVF International Conference on Computer Vision (ICCV), pages 15977–15987, 2023

    2023

  12. [20]

    DiffPose: Reliable 2D pose estimation through denoising diffusion

    Liu Jinwei, Zhang Feng, and Chen Lei. DiffPose: Reliable 2D pose estimation through denoising diffusion. InChina Automation Congress (CAC), pages 4041–4046. IEEE, 2023

    2023

  13. [21]

    What uncertainties do we need in bayesian deep learning for computer vision?Advances in neural information processing systems, 30, 2017

    Alex Kendall and Yarin Gal. What uncertainties do we need in bayesian deep learning for computer vision?Advances in neural information processing systems, 30, 2017

    2017

  14. [22]

    Hille, Ingrid Ullmann, Matthias Braun, et al

    Daniel Krauss, Lukas Engel, Tabea Ott, Johanna Br ¨aunig, Robert Richer, Markus Gambietz, Nils Albrecht, Eva M. Hille, Ingrid Ullmann, Matthias Braun, et al. A review and tutorial on machine learning-enabled radar-based biomedical monitoring.IEEE Open Journal of Engineering in...

    2024

  15. [23]

    HuPR: A benchmark for human pose estimation using millimeter wave radar

    Shih-Po Lee, Niraj Prakash Kini, Wen-Hsiao Peng, Ching- Wen Ma, and Jenq-Neng Hwang. HuPR: A benchmark for human pose estimation using millimeter wave radar. InPro- ceedings of the IEEE/CVF Winter Conference on Applica- tions of Computer Vision, pages 5715–5724, 2023

    2023

  16. [24]

    Generating multiple hypotheses for 3d human pose estimation with mixture density network

    Chen Li and Gim Hee Lee. Generating multiple hypotheses for 3d human pose estimation with mixture density network. InProceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 9887–9895, 2019

    2019

  17. [25]

    Human pose regression with 9 residual log-likelihood estimation

    Jiefeng Li, Siyuan Bian, Ailing Zeng, Can Wang, Bo Pang, Wentao Liu, and Cewu Lu. Human pose regression with 9 residual log-likelihood estimation. In2021 IEEE/CVF In- ternational Conference on Computer Vision (ICCV), pages 11005–11014, 2021

    2021

  18. [26]

    Human pose regression with residual log-likelihood estimation

    Jiefeng Li, Siyuan Bian, Ailing Zeng, Can Wang, Bo Pang, Wentao Liu, and Cewu Lu. Human pose regression with residual log-likelihood estimation. InProceedings of the IEEE/CVF international conference on computer vision, pages 11025–11034, 2021

    2021

  19. [27]

    Rpdiff: Multi- hypothesis human pose estimation based on mmwave radar

    Lanxin Li, Che Liu, and Tie Jun Cui. Rpdiff: Multi- hypothesis human pose estimation based on mmwave radar. In2024 IEEE MTT-S International Microwave Workshop Se- ries on Advanced Materials and Processes for RF and THz Applications (IMWS-AMP), pages 1–3. IEEE, 2024

    2024

  20. [28]

    Diff- HPE: Multi-hypothesis human pose estimation based on mmWave radar and diffusion framework.IEEE Sensors Journal, 2025

    Lanxin Li, Che Liu, Wenming Yu, and Tie Jun Cui. Diff- HPE: Multi-hypothesis human pose estimation based on mmWave radar and diffusion framework.IEEE Sensors Journal, 2025

    2025

  21. [29]

    Continuous normalizing flows for uncertainty- aware human pose estimation

    Shipeng Liu, Ziliang Xiong, Bastian Wandt, and Per-Erik Forss´en. Continuous normalizing flows for uncertainty- aware human pose estimation. InImage Analysis, pages 276–291, Cham, 2025. Springer Nature Switzerland

    2025

  22. [30]

    Saputra, Peijun Zhao, Yasin Almalioglu, Pedro P

    Chris Xiaoxuan Lu, Muhamad Risqi U. Saputra, Peijun Zhao, Yasin Almalioglu, Pedro P. B. De Gusmao, Chang- hao Chen, Ke Sun, Niki Trigoni, and Andrew Markham. milliEgo: Single-chip mmWave radar aided egomotion es- timation via deep sensor fusion. InProceedings of the 18th Confe...

    2020

  23. [31]

    Es- kofier

    Jonas Leo Mueller, Lukas Engel, Eva Dorschky, Daniel Krauss, Ingrid Ullmann, Martin V ossiek, and Bjoern M. Es- kofier. RadProPoser: A framework for human pose esti- mation with uncertainty quantification from raw radar data. arXiv preprint arXiv:2508.03578, 2025

    2025 arXiv

  24. [32]

    Adaptive biofeedback for digital physiotherapy using sakoe- chiba constrained pose matching

    Jonas Leo Mueller, Alexander Weiss, and Bjoern M Eskofier. Adaptive biofeedback for digital physiotherapy using sakoe- chiba constrained pose matching. InInternational Confer- ence on AI in Healthcare, pages 227–241. Springer, 2025

    2025

  25. [33]

    Obtaining well calibrated probabilities using bayesian binning

    Mahdi Pakdaman Naeini, Gregory Cooper, and Milos Hauskrecht. Obtaining well calibrated probabilities using bayesian binning. InProceedings of the AAAI conference on artificial intelligence, 2015

    2015

  26. [34]

    Masked autoregressive flow for density estimation.Advances in Neural Information Processing Systems, 30, 2017

    George Papamakarios, Theo Pavlakou, and Iain Murray. Masked autoregressive flow for density estimation.Advances in Neural Information Processing Systems, 30, 2017

    2017

  27. [35]

    Normalizing flows for probabilistic modeling and inference

    George Papamakarios, Eric Nalisnick, Danilo Jimenez Rezende, Shakir Mohamed, and Balaji Lakshminarayanan. Normalizing flows for probabilistic modeling and inference. Journal of Machine Learning Research, 22(57):1–64, 2021

    2021

  28. [36]

    Multi-hypothesis 3d human pose estimation metrics favor miscalibrated distributions.arXiv preprint arXiv:2210.11179, 2022

    Paweł A Pierzchlewicz, R James Cotton, Mohammad Bashiri, and Fabian H Sinz. Multi-hypothesis 3d human pose estimation metrics favor miscalibrated distributions.arXiv preprint arXiv:2210.11179, 2022

    2022 arXiv

  29. [37]

    Richards.Fundamentals of Radar Signal Process- ing

    Mark A. Richards.Fundamentals of Radar Signal Process- ing. McGraw-Hill, 2005

    2005

  30. [38]

    Christian Rupprecht, Iro Laina, Robert DiPietro, Maximil- ian Baust, Federico Tombari, Nassir Navab, and Gregory D. Hager. Learning in an uncertain world: Representing am- biguity through multiple hypotheses. InIEEE International Conference on Computer Vision (ICCV), pages 359...

    2017

  31. [39]

    Sensor-enabled safety systems for human-robot collaboration: A review.IEEE Sensors Jour- nal, 2024

    Constantin Scholz, Hoang-Long Cao, Emil Imrith, Nima Roshandel, Hamed Firouzipouyaei, Aleksander Burkiewicz, Milan Amighi, Sebastien Menet, Dylan Warawout Sisavath, Antonio Paolillo, et al. Sensor-enabled safety systems for human-robot collaboration: A review.IEEE Sensors Jour...

    2024

  32. [40]

    mm-Pose: Real-time human skeletal posture estimation using mmWave radars and CNNs.IEEE Sensors Journal, 2019

    Arindam Sengupta, Feng Jin, Renyuan Zhang, and Siyang Cao. mm-Pose: Real-time human skeletal posture estimation using mmWave radars and CNNs.IEEE Sensors Journal, 2019

    2019

  33. [41]

    mm-Pose: Real-time human skeletal posture estimation using mmWave radars and CNNs.IEEE Sensors Journal, 20 (17):10032–10044, 2020

    Arindam Sengupta, Feng Jin, Renyuan Zhang, and Siyang Cao. mm-Pose: Real-time human skeletal posture estimation using mmWave radars and CNNs.IEEE Sensors Journal, 20 (17):10032–10044, 2020

    2020

  34. [42]

    Diffusion-based 3D human pose estimation with multi- hypothesis aggregation

    Wenkang Shan, Zhenhua Liu, Xinfeng Zhang, Zhao Wang, Kai Han, Shanshe Wang, Siwei Ma, and Wen Gao. Diffusion-based 3D human pose estimation with multi- hypothesis aggregation. InIEEE/CVF International Confer- ence on Computer Vision (ICCV), pages 14761–14771, 2023

    2023

  35. [43]

    Denois- ing diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denois- ing diffusion implicit models. InInternational Conference on Learning Representations, 2021

    2021

  36. [44]

    Privacy-preserving fall detection with deep learn- ing on mmwave radar signal

    Yangfan Sun, Renlong Hang, Zhu Li, Mouqing Jin, and Kelvin Xu. Privacy-preserving fall detection with deep learn- ing on mmwave radar signal. In2019 IEEE Visual Commu- nications and Image Processing (VCIP), pages 1–4. IEEE, 2019

    2019

  37. [45]

    Gomez, Łukasz Kaiser, and Illia Polosukhin

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszko- reit, Llion Jones, Aidan N. Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need.Advances in Neural Information Processing Systems, 30, 2017

    2017

  38. [46]

    Probabilistic monocular 3D human pose estimation with normalizing flows

    Tom Wehrbein, Marco Rudolph, Bodo Rosenhahn, and Bastian Wandt. Probabilistic monocular 3D human pose estimation with normalizing flows. InIEEE/CVF Inter- national Conference on Computer Vision (ICCV), pages 11199–11208, 2021

    2021

  39. [47]

    Learning likelihoods with conditional normal- izing flows.arXiv preprint arXiv:1912.00042, 2019

    Christina Winkler, Daniel Worrall, Emiel Hoogeboom, and Max Welling. Learning likelihoods with conditional normal- izing flows.arXiv preprint arXiv:1912.00042, 2019

    1912 arXiv

  40. [48]

    mmMesh: To- wards 3D real-time dynamic human mesh construction us- ing millimeter-wave

    Hongfei Xue, Yan Ju, Chenglin Miao, Yijiang Wang, Shiyang Wang, Aidong Zhang, and Lu Su. mmMesh: To- wards 3D real-time dynamic human mesh construction us- ing millimeter-wave. InACM SIGMOBILE International Conference on Mobile Systems, Applications, and Services, 2021

    2021

  41. [49]

    MM-Fi: Multi-modal non-intrusive 4D human dataset for versatile wireless sensing

    Jianfei Yang, He Huang, Yunjiao Zhou, Xinyan Chen, Yue- cong Xu, Shenghai Yuan, Han Zou, Chris Xiaoxuan Lu, and Lihua Xie. MM-Fi: Multi-modal non-intrusive 4D human dataset for versatile wireless sensing. InAdvances in Neu- ral Information Processing Systems (NeurIPS) Datasets...

    2023

  42. [50]

    Freeman, Rahul Sukthankar, and Cristian Smin- chisescu

    Andrei Zanfir, Eduard Gabriel Bazavan, Hongyi Xu, William T. Freeman, Rahul Sukthankar, and Cristian Smin- chisescu. Weakly supervised 3d human pose and shape re- construction with normalizing flows. InComputer Vision – 10 ECCV 2020, pages 465–481, Cham, 2020. Springer Interna...

    2020

  43. [51]

    Normalizing flows are capable generative models.arXiv preprint arXiv:2412.06329, 2024

    Shuangfei Zhai, Ruixiang Zhang, Preetum Nakkiran, David Berthelot, Jiatao Gu, Huangjie Zheng, Tianrong Chen, Miguel Angel Bautista, Navdeep Jaitly, and Josh Susskind. Normalizing flows are capable generative models.arXiv preprint arXiv:2412.06329, 2024

    2024 arXiv

  44. [52]

    Point transformer

    Hengshuang Zhao, Li Jiang, Jiaya Jia, Philip HS Torr, and Vladlen Koltun. Point transformer. InProceedings of the IEEE/CVF international conference on computer vision, pages 16259–16268, 2021

    2021

  45. [53]

    Deep learning-based human pose estimation: A survey.ACM Computing Surveys, 56(1):1–37, 2023

    Ce Zheng, Wenhan Wu, Chen Chen, Taojiannan Yang, Si- jie Zhu, Ju Shen, Nasser Kehtarnavaz, and Mubarak Shah. Deep learning-based human pose estimation: A survey.ACM Computing Surveys, 56(1):1–37, 2023. 11 A. Prior Loss Derivation We test two formalizations of the prior loss. G...

    2023

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.