Pith. sign in

REVIEW 5 major objections 6 minor 39 references

MultiFormer: A Multi-Person Pose Estimation System Based on CSI and Attention Mechanism

T0 review · 5 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A WiFi-only network can estimate multi-person body poses, including wrists and elbows, from radio signals.

desk verdict The dual-token CSI-Transformer idea is genuinely new and the ablations are coherent, but the SOTA comparison is undermined by an undocumented, internally inconsistent baseline protocol, so the headline accuracy claims are not yet established. read the letter →

arxiv 2505.22555 v2 pith:YRCGNRSX submitted 2025-05-28 cs.CV eess.SP

Yanyi Qu , Haoyang Ma , Wenhui Xiong This is my paper
classification cs.CVeess.SP
keywords humanposeestimationChannelStateInformationWiFisensingTransformermulti-headself-attentionmulti-personpartaffinityfieldsmulti-stagefeaturefusion
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

MultiFormer is a WiFi-based system that estimates the 2D skeleton of one or more people from Channel State Information (CSI), the fine-grained radio measurements of how a WiFi signal reflects off a room and its occupants. The paper's claim is that a dual-token Transformer, which models subcarrier correlations and temporal dynamics in separate streams, combined with a three-stage heatmap refinement network, yields more accurate pose estimates than previous WiFi pose estimators, especially for the wrists and elbows. The authors report the highest Percentage of Correct Keypoints scores among compared methods on both a self-collected dataset (PCK@5 of 0.5209) and a public benchmark dataset (PCK@20 of 0.7225), and demonstrate that the system works in darkness and in two-person scenes. If correct, this means privacy-preserving, camera-free, device-free human pose tracking is feasible with commodity WiFi hardware.

What carries the argument

The load-bearing object is the Time-Frequency Dual-Dimensional Tokenization (TFDDT): raw CSI amplitude is upsampled in both time and frequency, then grouped into 64 frequency tokens, each spanning time and antennas for one subcarrier, and 64 temporal tokens, each spanning subcarriers and antennas at one time instant. Two parallel multi-head self-attention streams with independent parameters process the two token types, and a reconstruction layer reshapes the one-dimensional outputs into 36x36 feature maps. The second mechanism is the Multi-Stage Feature Fusion Network (MSFN), whose Pose-Attentive Perception Module computes channel and spatial attention from the previous stage's PCM/PAF heatmaps and multiplies them into the CSI features before the next heatmap decoder. This iterative refinement is what allows the system to associate keypoints across multiple people and to focus progressively on body regions rather than background clutter.

What would settle it

Run the original released code for every compared method on a single fixed train/test split of each dataset and recompute PCK; if any model's parameter count or FLOPs changes between the two reported tables, the numbers were drawn from different publications and the comparison is not controlled.

Watch

Extended reading notes

Core claim

The central discovery is that CSI amplitude, when organized into separate time tokens and frequency tokens and processed by parallel multi-head self-attention modules, carries enough information to reconstruct multi-person 2D skeletons. Unlike prior approaches that treat CSI as an image-like tensor, regress keypoint coordinates directly, or encode only pairwise keypoint distances, MultiFormer produces Part Confidence Maps and Part Affinity Fields as intermediate heatmaps and refines them across three decoding stages. A Pose-Attentive Perception Module re-weights the CSI features channel-wise and spatially using the previous stage's heatmap output, which the authors argue enforces anatomical constraints. On their own dataset the full model reaches PCK@5 of 0.5209 and PCK@20 of 0.8885, and on the public benchmark PCK@20 of 0.7225, with the largest gains on wrists and elbows.

Load-bearing premise

The headline accuracy comparisons assume that every prior model was trained and evaluated under one shared protocol on each dataset, a premise the paper never documents and its own tables appear to contradict; the system also assumes that the vision-based teacher's 8 FPS webcam labels are accurate ground truth for both training and evaluation.

Editorial extensions

If this is right

  • A CSI-only system can estimate multi-person skeletons without cameras, wearables, or specialized radar hardware, so pose tracking remains available in darkness and privacy-sensitive settings.
  • The largest accuracy jumps occur on wrists and elbows, so gesture recognition and hand-motion analysis are the most promising near-term applications.
  • Multi-stage heatmap refinement with attention to the previous stage's output offers a way to enforce anatomical consistency that direct regression and pairwise-distance methods lack.
  • The full model uses 11.93M parameters and 15.12G FLOPs, while smaller variants still beat comparable-parameter baselines, so the approach can run at modest compute budgets.
  • Because the pipeline outputs the same part-confidence maps and part-affinity fields used in vision-based pose estimation, existing pose-decoding and multi-person association algorithms can be reused directly.

Reading between the lines

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

  • The dual-token separation suggests that treating frequency and time as independent relational dimensions before fusing them may transfer to other CSI tasks, such as activity recognition or device-free localization, where similar tokenization could help.
  • The reported wrist and elbow gains could be probed more directly by evaluating on a larger public dataset with more subjects and multiple room layouts; if the margins persist across environments, the improvement is tied to the architecture rather than to the specific recording setup.
  • The teacher-student setup means the system inherits whatever biases the vision-based teacher has, so a natural next test is training from motion-capture labels to see whether the CSI estimator can exceed the teacher's precision.
  • A three-stage decoder with attention reweighting is a generic refinement recipe that could be grafted onto other sensor modalities, not just WiFi, wherever intermediate heatmaps are available.
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

5 major / 6 minor

Summary. The paper proposes MultiFormer, a CSI-based multi-person 2D pose estimation system. It introduces Time-Frequency Dual-Dimensional Tokenization (TFDDT), which converts upsampled CSI amplitude into separate temporal and frequency token streams processed by two parallel multi-head self-attention encoders; the extracted features are fed to a Multi-Stage Feature Fusion Network (MSFN) that iteratively refines PCM and PAF heatmaps using channel and spatial attention, and the final pose is decoded via NMS and Hungarian matching. Training labels are obtained from an OpenPose teacher on synchronized webcam video, and the student network is evaluated on a self-collected two-person dataset and on the public MM-Fi dataset. The paper reports consistent improvements in PCK over several baselines, with the largest gains on wrist and elbow keypoints, and claims state-of-the-art accuracy across PCK@5 to PCK@40.

Significance. If the reported results are reliable, the architecture is a plausible contribution to privacy-preserving, camera-free multi-person pose estimation: the dual-token transformer design is a clean way to capture both subcarrier and temporal structure, the multi-stage fusion with PAPM is well motivated, and the teacher-student label pipeline is disclosed rather than hidden. The ablation study is internally consistent and isolates single factors, which is a genuine strength. However, the current evidence does not yet support the abstract's headline claim of superiority over state-of-the-art methods: the baseline comparison protocol is undocumented and internally inconsistent, the self-collected evaluation uses only two volunteers and no error bars, and the MM-Fi skeleton is not reconciled with the model's 18-keypoint output. The engineering ideas are interesting and potentially publishable, but the experimental validation needs substantial rework.

major comments (5)
  1. [IV-C, Tables VII and VIII] The baseline comparison is not based on a documented shared protocol, and the tables contradict the assumption of comparability. WPFormer is listed as 30.81M parameters and 8.26G FLOPs in Table VII but as 26.42M parameters and 507.89G FLOPs in Table VIII; WISPPN is 20.19M/7.70G in Table VII but 26.78M/159.81G in Table VIII. A fixed architecture cannot have different parameter counts across tables, so these numbers must come from different model variants or different publications with different input resolutions and evaluation settings. The paper never states the MM-Fi train/test split, whether baselines were retrained or imported, the input size used for each baseline, or how PCK normalization was applied. This makes the headline MM-Fi ranking, including the 0.7225 versus 0.6841 PCK@20 lead over CSI2Pose, unsupported as a head-to-head comparison.
  2. [IV-A and IV-C, Eqs. (13)-(15), Fig. 15] Teacher labels are generated by OpenPose from an approximately 8 FPS webcam and used as ground truth for both training and evaluation of the student. The paper does not describe how the 50 Hz CSI frames are synchronized with 8 FPS video, nor how frames in which OpenPose fails (or produces uncertain keypoints) are handled. Because the evaluation uses the same teacher outputs used for training, the self-collected PCK measures agreement with the teacher rather than physical pose accuracy. This is especially relevant to the claimed robustness in darkness in Section IV-C, since the student is trained on OpenPose labels rather than on independently verified pose data.
  3. [Table I, Section IV-A, Figs. 13 and 14] MM-Fi is described as containing 17-keypoint pose annotations, while MultiFormer outputs an 18-keypoint OpenPose-style skeleton with a 19-channel PCM and 38-channel PAF representation. The paper never states how MM-Fi's 17 keypoints are mapped to the 18-keypoint output, how the teacher PCM/PAF labels are generated on MM-Fi, or how the 17-keypoint results in Figs. 13 and 14 are derived when the model produces 18 keypoints. The keypoint names in Fig. 13 (e.g., Bot Torso, Center Torso) are not the standard OpenPose/COCO 18-keypoint names, which suggests a separate label convention. Without this mapping, the MM-Fi results in Table VIII and Figs. 13-14 cannot be audited.
  4. [IV-A, Eq. (16)] The PCK definition in Eq. (16) is ambiguous or incorrect as written. The denominator is sqrt(rs^2 + lh^2), where rs and lh are described as the 'positions' of the right shoulder and left hip. A standard PCK normalizes the keypoint error by a scale derived from the torso or head segment length, not by the Euclidean norm of two coordinate vectors, so Eq. (16) is not a standard PCK and its normalization is unclear. Since all reported results are PCK values, this issue affects every table and the cross-dataset comparability of Tables VII and VIII.
  5. [IV-A, IV-B, Tables III-VIII] Every ablation and comparison table reports a single run with no error bars or cross-validation, and the self-collected dataset has only two volunteers (Table I). The central quantitative claims, such as the 11% improvement from dual-token over single-token modeling (Table III) and the 12%-64% stage-wise improvements in Table IV, would be strengthened substantially by reporting mean and standard deviation over multiple runs or subject-wise cross-validation. As reported, the differences could be within run-to-run or subject-dependent noise. This does not invalidate the architecture, but it limits confidence in the magnitude of the claimed gains.
minor comments (6)
  1. [Abstract and Section III-C] The abstract contains a duplicated word: 'Extensive experiments conducted on on the public MM-Fi dataset'.
  2. [Section III-C and Fig. 7] The text has another duplication: 'at at different stages' in Section III-C and the caption of Fig. 7.
  3. [Table VI] Several values in Table VI are missing decimal points or formatting, e.g., '02809', '02603', '02550', which should be '0.2809', '0.2603', '0.2550'.
  4. [Figures 3, 4, and 8] The text and captions for Figures 3, 4, and 8 contain unresolved glyph codes such as '/uni00000013' instead of readable labels; the figures need to be regenerated cleanly before publication.
  5. [References] Reference [27], cited for CSI2Pose, is titled 'Towards cross-modal forgery detection and localization on live surveillance videos', which does not match the CSI-pose-estimation method described in the text; the citation should be checked and corrected.
  6. [Throughout] The spelling of the WiSPPN baseline is inconsistent: the text uses both 'WISPPN' and 'WiSPPN', and Table VI uses 'WISPPN' while Table VII uses 'WiSPPN'. Please unify the notation.

Circularity Check

0 steps flagged · score 1.0 of 10

No construction-level circularity found; the disclosed OpenPose teacher-student label loop is a benchmark limitation, not a circular derivation.

full rationale

The derivation chain uses standard definitions (TFDDT tokenization Eqs. 1-4; multi-head attention Eq. 6; MSFN/PAPM Eqs. 7-9; PAF-integral/Hungarian matching Eqs. 10-12; PCK Eq. 16), with no fitted parameter renamed as a prediction. The one dependence is the teacher-student setup: OpenPose-derived PCM/PAF are the training targets (Eqs. 13-15), and on the self-collected track the same teacher outputs serve as the PCK ground truth (Eq. 16). This is explicitly disclosed in Section III and is standard for distillation; it means the self-collected accuracy is a measure of fidelity to the OpenPose teacher, but it does not make the CSI-to-pose mapping equivalent to its inputs by construction, since different architectures obtain different PCK values and the model must learn a nontrivial cross-modal mapping. The MM-Fi evaluation uses an external dataset with its own annotations, giving independent support. The WPFormer parameter/FLOP discrepancy between Tables VII and VIII and the undocumented baseline protocol are correctness/auditability concerns, not circularity. No load-bearing self-citations or imported uniqueness theorems appear.

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

The central claims rest on standard deep-learning practice: a teacher-student training pipeline, hand-chosen architecture sizes, and evaluation against the same OpenPose labels used for training. No new physical entities are introduced. The main unpaid-for inputs are the accuracy of the OpenPose-derived labels and the unspecified adaptation of TFDDT to MM-Fi's 114-subcarrier layout, which the Section III-A description does not cover.

free parameters (5)
  • Number of MSFN decoding stages = 3
    Chosen by hand and validated by ablation (Table IV) showing PCK improves from stage 1 to stage 3, but no stage-4 or cost-accuracy analysis is given; the headline accuracies depend on this choice.
  • Token sequence size per stream = 64 tokens x 1296
    Described in Section III-A as balancing computation and accuracy; no sensitivity analysis is provided.
  • Upsampled CSI dimensions = 64 x 3 x 64
    Result of zero-insertion and low-pass upsampling from 10 x 3 x 30 (Eq. 2); the upsample ratio N and the filter g_T are not specified, and the target size is a design choice affecting all results.
  • Weight decay = 0.7
    Table II lists weight decay 0.7, an unusual value for L2 regularization; no justification or sensitivity analysis is given.
  • Number of attention heads H = not reported
    Eq. (6) defines multi-head attention with H heads, but H is never stated, leaving the architecture underspecified.
assumptions (4)
  • domain assumption OpenPose-generated PCM and PAF labels are accurate enough to serve as ground truth for training and evaluation.
    The teacher network labels the self-collected data from an 8 FPS webcam (Section III and Section IV-A); motion blur, occlusion, and low frame rate in these labels propagate directly into the reported PCK numbers.
  • domain assumption CSI amplitude variations encode pose-dependent multipath structure that a Transformer can map to body keypoints.
    This is the system premise stated in Section I; the paper provides no independent verification that amplitude-only features carry pose information rather than environment-specific artifacts.
  • domain assumption The low-pass interpolation of Eq. (2) does not distort pose-relevant CSI structure.
    Upsampling inserts zeros and low-pass filters in time and frequency (Section III-A); the filter is not specified and its effect on pose discriminability is not validated.
  • ad hoc to paper MM-Fi's 17-keypoint skeleton is compatible with the 18-keypoint OpenPose PCM/PAF representation.
    Section III-C describes 19 PCM channels for 18 keypoints plus background, while MM-Fi provides 17-keypoint annotations; the mapping between skeletons is never described, yet Table VIII and Figs. 13-14 report per-keypoint accuracy on MM-Fi.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MultiFormer: A Multi-Person Pose Estimation System Based on CSI and Attention Mechanism." pith.science (2026). https://pith.science/paper/YRCGNRSX

@misc{pith2026250522555,
  author       = {Pith},
  title        = {Pith review of: MultiFormer: A Multi-Person Pose Estimation System Based on CSI and Attention Mechanism},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YRCGNRSX}},
  note         = {Machine review of arXiv:2505.22555}
}
read the original abstract

Human pose estimation based on Channel State Information (CSI) has emerged as a promising approach for non-intrusive and precise human activity monitoring, yet faces challenges including accurate multi-person pose recognition and effective CSI feature learning. This paper presents MultiFormer, a wireless sensing system that accurately estimates human pose through CSI. The proposed system adopts a Transformer based time-frequency dual-token feature extractor with multi-head self-attention. This feature extractor is able to model inter-subcarrier correlations and temporal dependencies of the CSI. The extracted CSI features and the pose probability heatmaps are then fused by Multi-Stage Feature Fusion Network (MSFN) to enforce the anatomical constraints. Extensive experiments conducted on on the public MM-Fi dataset and our self-collected dataset show that the MultiFormer achieves higher accuracy over state-of-the-art approaches, especially for high-mobility keypoints (wrists, elbows) that are particularly difficult for previous methods to accurately estimate.

Figures

Figures reproduced from arXiv: 2505.22555 by the authors.

Figure 1
Figure 1. Block Diagram Of MultiFormer fewer parameter count and computational cost, giving better efficiency and performance. C. Human Pose Estimation Techniques Accurately estimating the positions of skeletal keypoints is crucial in human pose estimation. Current methods for this task can be categorized into direct regression, PAM regression, and heatmap-based approaches. Wang et al. [25] proposed the Wi-Mose model, which u… view at source ↗
Figure 2
Figure 2. Workflow of Time-Frequency Dual-Dimensional Tokenization [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Block diagram of multi-head self-attention feature extraction module [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (13 more)
Figure 4
Figure 4. Figure 4: Key subcarriers and key intervals of multi-head self-attention module for different movements. (a) key subcarriers for ’Push’ movement, (b) key time [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 6
Figure 6. Figure 6: The block diagram of Heatmap Decoder The Heatmap Decoder estimates PCM and PAF heatmaps from encoded CSI features by CNN to capture the spatial structure from the feature. Hi = {Pi , Ai} = Decoder(Φi ;WD) (9) [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 5
Figure 5. Figure 5: The block diagram of Pose-Attentive Perception Module (PAPM) [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 7
Figure 7. Figure 7: Pose probability estimation results and decoded poses of Multi-Stage Pose Feature Fusion Network (MSFN) at at different stages. (a) stage 1, (b) [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: The Channel Attention weights of Multi-Stage Pose Feature Fusion [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Spatial Weights from Multi-Stage Pose Feature Fusion Network [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 10
Figure 10. Figure 10: Top-down view of the data collection setup. (a) Location 1, (b) [PITH_FULL_IMAGE:figures/full_fig_p008_10.png]
Figure 11
Figure 11. Figure 11: Comparison of pose estimation results between the PAM-based method (up) and MultiFormer (down) on single-person poses for 8 movement types [PITH_FULL_IMAGE:figures/full_fig_p009_11.png]
Figure 12
Figure 12. Figure 12: Comparison of pose estimation results between the PAM-based [PITH_FULL_IMAGE:figures/full_fig_p010_12.png]
Figure 14
Figure 14. Figure 14: Comparison of the PCK@20 for 17 keypoints for HPE-Li and [PITH_FULL_IMAGE:figures/full_fig_p011_14.png]
Figure 13
Figure 13. Figure 13: Comparison of the PCK@20 for 17 keypoints for CSI2Pose and [PITH_FULL_IMAGE:figures/full_fig_p011_13.png]
Figure 15
Figure 15. Figure 15: Comparison of pose estimation between OpenPose that fails to [PITH_FULL_IMAGE:figures/full_fig_p011_15.png]
Figure 16
Figure 16. Figure 16: Comparison of pose estimation between OpenPose that fails to [PITH_FULL_IMAGE:figures/full_fig_p012_16.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

39 extracted references · 35 canonical work pages

  1. [1]

    Sensor-based and vision-based human activity recognition: A comprehensive survey,

    L. Minh Dang, K. Min, H. Wang, Md. Jalil Piran, C. Hee Lee, and H. Moon, “Sensor-based and vision-based human activity recognition: A comprehensive survey,”Pattern Recognition, vol. 108, p. 107561, Dec. 2020

    work page 2020

  2. [2]

    Wireless Sensing for Human Activity: A Survey,

    J. Liu, H. Liu, Y . Chen, Y . Wang, and C. Wang, “Wireless Sensing for Human Activity: A Survey,”IEEE Communications Surveys & Tutorials, vol. 22, no. 3, pp. 1629–1645, 2020

    work page 2020

  3. [3]

    A survey on video-based Human Action Recognition: Recent updates, datasets, challenges, and applications,

    P. Pareek and A. Thakkar, “A survey on video-based Human Action Recognition: Recent updates, datasets, challenges, and applications,” Artificial Intelligence Review, vol. 54, pp. 2259–2322, Mar. 2021

    work page 2021

  4. [4]

    Human activity recognition in artificial intelligence framework: A narrative review,

    N. Gupta, S. K. Gupta, R. K. Pathak, V . Jain, P. Rashidi, and J. S. Suri, “Human activity recognition in artificial intelligence framework: A narrative review,”Artificial Intelligence Review, vol. 55, pp. 4755– 4808, Aug. 2022

    work page 2022

  5. [5]

    Openpose: Re- altime multi-person 2d pose estimation using part affinity fields,

    Z. Cao, G. Hidalgo, T. Simon, S.-E. Wei, and Y . Sheikh, “Openpose: Re- altime multi-person 2d pose estimation using part affinity fields,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, pp. 172–186, Jan. 2021

    work page 2021

  6. [6]

    Integrating 2d and 3d human pose estimation using depth camera in cardio exercises,

    A. K. Wei Tiong, K. H. Lim, J. T. Sien Phang, and P. K. Pang, “Integrating 2d and 3d human pose estimation using depth camera in cardio exercises,” in2024 International Conference on Green Energy, Computing and Sustainable Technology (GECOST), pp. 255–259, 2024

    work page 2024

  7. [7]

    Lwirpose: A novel long wave infrared thermal image pose dataset and benchmark,

    A. Upadhyay, B. Dhupar, M. Sharma, A. Shukla, and A. Abraham, “Lwirpose: A novel long wave infrared thermal image pose dataset and benchmark,” in2024 IEEE International Conference on Image Processing (ICIP), pp. 186–192, 2024

    work page 2024

  8. [8]

    Structured feature learning for pose estimation,

    X. Chu, W. Ouyang, H. Li, and X. Wang, “Structured feature learning for pose estimation,” in2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (Las Vegas, NV , USA), pp. 4715–4723, IEEE, June 2016

    work page 2016

Show all 39 references
  1. [9]

    Rmpe: Regional multi-person pose estimation,

    H.-S. Fang, S. Xie, Y .-W. Tai, and C. Lu, “Rmpe: Regional multi-person pose estimation,” in2017 IEEE International Conference on Computer Vision (ICCV), (Venice), pp. 2353–2362, IEEE, Oct. 2017

    2017

  2. [10]

    Benchmarking and error diagnosis in multi-instance pose estimation,

    M. R. Ronchi and P. Perona, “Benchmarking and error diagnosis in multi-instance pose estimation,” in2017 IEEE International Conference on Computer Vision (ICCV), (Venice), pp. 369–378, IEEE, Oct. 2017

    2017

  3. [11]

    Thin-slicing network: A deep structured model for pose estimation in videos,

    J. Song, L. Wang, L. Van Gool, and O. Hilliges, “Thin-slicing network: A deep structured model for pose estimation in videos,” in2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (Honolulu, HI), pp. 5563–5572, IEEE, July 2017

    2017

  4. [12]

    Confidence-based 6d object pose estimation,

    W.-L. Huang, C.-Y . Hung, and I.-C. Lin, “Confidence-based 6d object pose estimation,”IEEE Transactions on Multimedia, vol. 24, pp. 3025– 3035, 2022

    2022

  5. [13]

    A contactless health monitoring system for vital signs monitoring, human activity recognition, and tracking,

    A. Li, E. Bodanese, S. Poslad, P. Chen, J. Wang, Y . Fan, and T. Hou, “A contactless health monitoring system for vital signs monitoring, human activity recognition, and tracking,”IEEE Internet of Things Journal, vol. 11, no. 18, pp. 29275–29286, 2024

    2024

  6. [14]

    Body-worn imu human skeletal pose estimation using a factor graph-based optimization framework,

    T. McGrath and L. Stirling, “Body-worn imu human skeletal pose estimation using a factor graph-based optimization framework,”Sensors, vol. 20, p. 6887, Dec. 2020

    2020

  7. [15]

    The accuracy and precision of position and orientation tracking in the htc vive virtual reality system for scientific research,

    D. C. Niehorster, L. Li, and M. Lappe, “The accuracy and precision of position and orientation tracking in the htc vive virtual reality system for scientific research,”i-Perception, vol. 8, p. 2041669517708205, June 2017

    2017

  8. [16]

    Low-cost and device- free human activity recognition based on hierarchical learning model,

    J. Chen, X. Huang, H. Jiang, and X. Miao, “Low-cost and device- free human activity recognition based on hierarchical learning model,” Sensors, vol. 21, p. 2359, Mar. 2021

    2021

  9. [17]

    From RSSI to CSI: Indoor localization via channel response,

    Z. Yang, Z. Zhou, and Y . Liu, “From RSSI to CSI: Indoor localization via channel response,”ACM Computing Surveys, vol. 46, pp. 1–32, Nov. 2013

    2013

  10. [18]

    Attention-Based Gesture Recognition Using Commodity WiFi De- vices,

    Y . Gu, H. Yan, X. Zhang, Y . Wang, J. Huang, Y . Ji, and F. Ren, “Attention-Based Gesture Recognition Using Commodity WiFi De- vices,”IEEE Sensors Journal, vol. 23, pp. 9685–9696, May 2023

    2023

  11. [19]

    Metafi: Device-free pose estimation via commodity wifi for metaverse avatar simulation,

    J. Yang, Y . Zhou, H. Huang, H. Zou, and L. Xie, “Metafi: Device-free pose estimation via commodity wifi for metaverse avatar simulation,” in2022 IEEE 8th World Forum on Internet of Things (WF-IoT), (Yokohama, Japan), pp. 1–6, IEEE, Oct. 2022

    2022

  12. [20]

    MetaFi++: WiFi-Enabled Transformer-Based Human Pose Estimation for Metaverse Avatar Simulation,

    Y . Zhou, H. Huang, S. Yuan, H. Zou, L. Xie, and J. Yang, “MetaFi++: WiFi-Enabled Transformer-Based Human Pose Estimation for Metaverse Avatar Simulation,”IEEE Internet of Things Journal, vol. 10, pp. 14128– 14136, Aug. 2023

    2023

  13. [21]

    Towards 3d human pose construction using wifi,

    W. Jiang, H. Xue, C. Miao, S. Wang, S. Lin, C. Tian, S. Murali, H. Hu, Z. Sun, and L. Su, “Towards 3d human pose construction using wifi,” inProceedings of the 26th Annual International Conference on Mobile Computing and Networking, (London United Kingdom), pp. 1– 14, ACM, Apr. 2020

    2020

  14. [22]

    CSI-Former: Pay More Attention to Pose Estimation with WiFi,

    Y . Zhou, C. Xu, L. Zhao, A. Zhu, F. Hu, and Y . Li, “CSI-Former: Pay More Attention to Pose Estimation with WiFi,”Entropy, vol. 25, p. 20, Dec. 2022

    2022

  15. [23]

    Can WiFi Estimate Person Pose?

    F. Wang, S. Panev, Z. Dai, J. Han, and D. Huang, “Can WiFi Estimate Person Pose?.” arXiv preprint arXiv:1904.00277, Apr. 2019

    1904 arXiv

  16. [24]

    CHA-Sens: An End- to-End Comprehensive Residual Convolution Framework for CSI-based Human Activity Sensing,

    F. Zhou, W. Zhang, G. Zhu, H. Li, and Q. Shi, “CHA-Sens: An End- to-End Comprehensive Residual Convolution Framework for CSI-based Human Activity Sensing,” in2023 26th International Conference on Computer Supported Cooperative Work in Design (CSCWD), pp. 113– 118, May 2023

    2023

  17. [25]

    From Point to Space: 3D Moving Human Pose Estimation Using Commodity WiFi,

    Y . Wang, L. Guo, Z. Lu, X. Wen, S. Zhou, and W. Meng, “From Point to Space: 3D Moving Human Pose Estimation Using Commodity WiFi,” IEEE Communications Letters, vol. 25, pp. 2235–2239, July 2021

    2021

  18. [26]

    Person-in- wifi: Fine-grained person perception using wifi,

    F. Wang, S. Zhou, S. Panev, J. Han, and D. Huang, “Person-in- wifi: Fine-grained person perception using wifi,” in2019 IEEE/CVF International Conference on Computer Vision (ICCV), (Seoul, Korea (South)), pp. 5451–5460, IEEE, Oct. 2019

    2019

  19. [27]

    Towards cross- modal forgery detection and localization on live surveillance videos,

    Y . Huang, X. Li, W. Wang, T. Jiang, and Q. Zhang, “Towards cross- modal forgery detection and localization on live surveillance videos,” IEEE INFOCOM, pp. 1–10, 2021

    2021

  20. [28]

    CSI-F: A Human Motion Recognition Method Based on Channel-State-Information Signal Feature Fusion,

    J. Niu, X. He, B. Fang, G. Han, X. Wang, and J. He, “CSI-F: A Human Motion Recognition Method Based on Channel-State-Information Signal Feature Fusion,”Sensors, vol. 24, p. 862, Jan. 2024

    2024

  21. [29]

    Person-in-wifi 3d: End-to-end multi-person 3d pose estimation with wi-fi,

    K. Yan, F. Wang, B. Qian, H. Ding, J. Han, and X. Wei, “Person-in-wifi 3d: End-to-end multi-person 3d pose estimation with wi-fi,” in2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), (Seattle, W A, USA), pp. 969–978, IEEE, June 2024

    2024

  22. [30]

    MDPose: Human Skeletal Motion Reconstruction Using WiFi Micro- Doppler Signatures,

    C. Tang, W. Li, S. Vishwakarma, F. Shi, S. Julier, and K. Chetty, “MDPose: Human Skeletal Motion Reconstruction Using WiFi Micro- Doppler Signatures,”IEEE Transactions on Aerospace and Electronic Systems, vol. 60, pp. 157–167, Feb. 2024

    2024

  23. [31]

    Robust wifi sensing-based human pose estimation using denoising autoencoder and cnn with dynamic subcarrier attention,

    X. H. Nguyen, V .-D. Nguyen, Q.-T. Luu, T. D. Gian, and O.-S. Shin, “Robust wifi sensing-based human pose estimation using denoising autoencoder and cnn with dynamic subcarrier attention,”IEEE Internet of Things Journal, pp. 1–1, 2025. IEEE Journal, VOL. XX, NO. XX, XXXX 2024 13

    2025

  24. [32]

    WiTransformer: A Novel Robust Gesture Recognition Sensing Model with WiFi,

    M. Yang, H. Zhu, R. Zhu, F. Wu, L. Yin, and Y . Yang, “WiTransformer: A Novel Robust Gesture Recognition Sensing Model with WiFi,” Sensors, vol. 23, p. 2612, Feb. 2023

    2023

  25. [33]

    Vision Transformers for Human Activity Recognition Using WiFi Channel State Information,

    F. Luo, S. Khan, B. Jiang, and K. Wu, “Vision Transformers for Human Activity Recognition Using WiFi Channel State Information,”IEEE Internet of Things Journal, vol. 11, pp. 28111–28122, Sept. 2024

    2024

  26. [34]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. u. Kaiser, and I. Polosukhin, “Attention is all you need,” inAdvances in Neural Information Processing Systems(I. Guyon, U. V . Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett...

    2017

  27. [35]

    Mm-fi: Multi-modal non-intrusive 4d human dataset for versatile wireless sensing,

    J. Yang, H. Huang, Y . Zhou, X. Chen, Y . Xu, S. Yuan, H. Zou, C. X. Lu, and L. Xie, “Mm-fi: Multi-modal non-intrusive 4d human dataset for versatile wireless sensing,”Advances in Neural Information Processing Systems, vol. 36, pp. 18756–18768, 2023

    2023

  28. [36]

    2d human pose estimation: New benchmark and state of the art analysis,

    M. Andriluka, L. Pishchulin, P. Gehler, and B. Schiele, “2d human pose estimation: New benchmark and state of the art analysis,” in2014 IEEE Conference on Computer Vision and Pattern Recognition, (Columbus, OH, USA), pp. 3686–3693, IEEE, June 2014

    2014

  29. [37]

    Hpe-li: Wifi-enabled lightweight dual selective kernel con- volution for human pose estimation,

    T. D. Gian, T. Dac Lai, T. Van Luong, K.-S. Wong, and V .-D. Nguyen, “Hpe-li: Wifi-enabled lightweight dual selective kernel con- volution for human pose estimation,” inComputer Vision – ECCV 2024(A. Leonardis, E. Ricci, S. Roth, O. Russakovsky, T. Sattler, and G. Varol, eds.)...

    2024

  30. [38]

    Powerskel: A device-free framework using csi signal for human skeleton estimation in power station,

    C. Yin, X. Miao, J. Chen, H. Jiang, J. Yang, Y . Zhou, M. Wu, and Z. Chen, “Powerskel: A device-free framework using csi signal for human skeleton estimation in power station,”IEEE Internet of Things Journal, vol. 11, no. 11, pp. 20165–20177, 2024

    2024

  31. [39]

    Towards robust and realistic human pose estimation via wifi signals,

    Y . Chen, J. Guo, S. Guo, J. Zhou, and D. Tao, “Towards robust and realistic human pose estimation via wifi signals,”arXiv preprint arXiv:2501.09411, 2025. Yanyi QuUndergraduate student at the School of Information and Communi- cation Engineering, University of Electronic Scie...

    2025 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.