Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

P2P-Insole: Human Pose Estimation Using Foot Pressure Distribution and Motion Sensors

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read A sub-dollar insole plus ankle motion can reconstruct a 21-joint 3D skeleton.

desk verdict Plausible low-cost insole pose-estimation hardware, but the current evaluation has a temporal leakage problem and no baseline comparison, so the accuracy claims are not yet established. read the letter →

arxiv 2505.00755 v1 pith:E2FZBGSN submitted 2025-05-01 cs.CV cs.AI

Atsuya Watanabe , Ratna Aisuwarya , Lei Jing This is my paper · ORCID
classification cs.CVcs.AI
keywords 3DhumanposeestimationinsolepressuresensorfootdistributionwearableIMUTransformere-textileprivacy-preservingsensingskeletonprediction
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

P2P-Insole claims that a pair of embroidered insole sensors, each with 35 pressure points and an ankle-mounted IMU, carries enough information to reconstruct a 21-joint 3D human skeleton across standing, tilting, bowing, sit-to-stand, and squatting. The system reports root-mean-square joint errors between 48.9 mm and 75.2 mm on those tasks, with the largest errors in squatting and the smallest in tilting. If this holds, full-body pose estimation no longer requires cameras, controlled lighting, or expensive motion-capture suits, which matters for rehabilitation, injury prevention, and continuous health monitoring in private spaces. The paper also reports that a Transformer trained on pressure, acceleration, rotation, and their first and second derivatives outperforms an LSTM baseline, with derivatives helping most on large-movement tasks.

What carries the argument

The load-bearing machinery is an 8-layer Transformer encoder over time-series frames, where each input frame concatenates 35 pressure values, 3 acceleration axes, and 3 rotation axes from each foot, together with the first and second derivatives of those signals. The derivatives turn static pressure images into motion-sensitive temporal features, which is what lets a foot-only sensor infer upper-body joints such as the head and arms. The 35-sensor insole geometry, the 100 Hz sampling, and the ankle IMU placement together define the input stream; the paper argues that this stream, not the model alone, is what makes sub-8 cm full-body estimates possible.

What would settle it

Retrain the same architecture on the same data using leave-one-participant-out cross-validation or a time-contiguous split that places whole movement blocks in either training or validation; if RMSE rises well above the reported 75 mm, the central claim of generalizable foot-to-pose estimation fails.

Watch

Extended reading notes

Core claim

The paper claims that plantar pressure distribution plus foot acceleration and rotation is a sufficient signal for estimating full-body 3D pose in a set of daily movements. A custom e-textile insole (under USD 1 per unit, with 35 piezoresistive Velostat-based sensors) captures pressure at 100 Hz, an integrated IMU adds 3-axis acceleration and rotation, and an 8-layer Transformer with 8 attention heads maps an 82-dimensional per-frame input to a 63-dimensional skeleton output. The model is trained on eighty minutes of synchronized data from four participants wearing OptiTrack suits, with first and second time-derivatives of each input channel appended. Reported RMSE is 48.9 mm for tilting, 52.6 mm for standing, 63.7 mm for bowing, 70.1 mm for standing and sitting, and 75.2 mm for squatting, and the Transformer outperforms an LSTM baseline across all tasks.

Load-bearing premise

The load-bearing assumption is that a random 8:2 split of consecutive frames from four people's movement sessions measures true accuracy; if neighboring frames from the same movement appear in both training and validation, the model may memorize pressure patterns instead of learning a general foot-to-skeleton mapping.

Editorial extensions

If this is right

  • A camera-free pose-estimation pipeline becomes feasible in bedrooms, bathrooms, and workplaces where cameras are unacceptable.
  • The reported error distribution implies that lower-body joints are estimated much more tightly than the head and arms; spine median error is about 6 mm while head median error is about 56 mm.
  • Adding time derivatives to the input is beneficial for large-amplitude motions such as bowing and squatting but can hurt small-movement accuracy such as standing and tilting, so input augmentation should be task-dependent.
  • Because the insole costs under USD 1 and uses an off-the-shelf microcontroller, the hardware could be mass-produced for continuous daily monitoring rather than lab sessions.

Reading between the lines

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

  • The reported accuracy is measured on a random 8:2 frame split from four participants; a stricter leave-one-participant-out evaluation would reveal how much of the learned mapping is person-specific and how much generalizes to new users.
  • The derivative trade-off suggests a task-adaptive input selector: include derivatives for dynamic movements and exclude them for quasi-static poses, a design the paper does not test.
  • If the foot-signal mapping generalizes, the same input stream could support contact-based fall prediction and gait analysis without a skeleton label, since the Transformer's temporal features already encode balance and weight shift.
  • Sensor-count reduction based on the reported sensor-position/error relationship could identify a minimal subset of pressure points that preserves accuracy, which the conclusions flag as future work.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes P2P-Insole, a system that estimates a 21-joint 3D skeleton from foot pressure distribution and motion sensors. The hardware consists of a sub-dollar e-textile insole with 35 pressure sensors and an ankle-mounted IMU; the learning model is an 8-layer Transformer with 8 attention heads trained with MSE loss on synchronized OptiTrack ground truth, pressure, and IMU data. The input stream is augmented with first and second derivatives. Experiments on four participants performing daily movements report RMSE between 48.9 and 75.2 mm across five tasks, plus ablations comparing the Transformer against LSTM and with/without derivative features. The paper claims that the system enables privacy-preserving, camera-free pose estimation for rehabilitation, injury prevention, and health monitoring.

Significance. If the reported accuracy were established, the system would be a practically significant contribution: the sub-dollar hardware cost, the fusion of pressure and IMU data, and the Transformer-with-derivatives design are all reasonable engineering choices, and the ablation against LSTM addresses a relevant design question. The paper also gives a detailed description of the sensor fabrication and data collection pipeline, which is useful for replication. However, the empirical evidence is currently too weak to support the central generalization claim: the evaluation uses only four participants, a random frame-level split, no subject-independent protocol, and no comparison against the directly related SolePoser system that the paper itself lists as achieving sub-70 mm error. No code or data release is mentioned, further limiting reproducibility. The strengths are the low-cost hardware and the clearly stated system architecture; the weakness is that the headline error numbers are not established as representative of new people or of the full daily-activity claim.

major comments (4)
  1. [§V-D and §IV (training protocol)] The 8:2 random frame-level split described in Section V-D, with training details in Section IV, is not a valid evaluation protocol for 100 Hz time-series data. Because each participant performed continuous movement repetitions and free movement in long sessions, random frame assignment places validation frames temporally adjacent to training frames from the same movement, with nearly identical pressure and IMU inputs and nearly identical skeleton labels. The model can therefore memorize local input-output mappings, and the reported RMSE values in Table II do not estimate performance on new people or new movement sequences. Please report a split by recording session or by participant (leave-one-subject-out), and report per-subject variability.
  2. [§V-A, §V-B, and Table II] The dataset consists of only four participants, and although eight movement classes are collected, Table II reports results only for Stand, Tilt, Bow, Stand and Sit, and Squat; walking, jumping, hopping, and standing on one leg are not evaluated. Without subject-independent evaluation, or at least per-participant results and variance estimates across random seeds, the abstract's claim of robustness and the conclusion's generalization claim are unsupported. The data-collection arithmetic is also inconsistent: eight movements times two minutes plus six minutes of free movement equals 22 minutes, not the stated 20 minutes per participant.
  3. [Table I and reference [6]] Table I lists SolePoser as using a similar 32-point insole with an IMU and a two-stream Transformer with sub-70 mm error, and reference [6] is cited, but no quantitative comparison, common benchmark, or discussion is provided. Since the paper's main claim is about pose-estimation accuracy, the lack of comparison against the closest prior art makes it impossible to determine what the new contribution achieves relative to the state of the art. Please add a same-protocol comparison or, if unavailable, clearly state why the numbers are not directly comparable and provide a task-matched baseline.
  4. [Figures 7 and 8 and Section V-D] The ablations in Figures 7 and 8 report single mean RMSE values without error bars, per-seed variance, or statistical tests. The statement in Section V-D that the Transformer model significantly improved performance, and the derivative trade-off conclusion, are therefore not supported by the evidence shown. Please report mean plus/minus standard deviation over multiple training runs or use paired tests over subjects.
minor comments (5)
  1. [Equation (1)] Equation (1) is described as a voltage divider circuit, but the expression is the standard non-inverting amplifier gain; please reconcile the circuit description with the equation.
  2. [Section IV (input notation)] The notation X in R^{N x d x 2} is unclear: if d is the dimension of the concatenated features, the factor of 2 and the role of N as the number of frames should be defined precisely, especially after the first and second derivatives are added.
  3. [Section V-C (synchronization)] The preprocessing section states that all data sources were aligned in terms of starting points, ending points, and total number of frames; please state whether data segments were trimmed or padded and how much data was discarded during synchronization.
  4. [Section V-A (IMU placement)] The manuscript says the IMU is attached to the subject's ankle but does not specify whether one or both ankles are instrumented; the input description of 82 total dimensions implies two feet, so please clarify the sensor placement.
  5. [Figures 7 and 8 (readability)] The axis labels and legends in Figures 7 and 8 appear partially machine-generated and are difficult to read; please redraw these figures for publication clarity.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the pose estimates are produced by supervised training on independent sensor and skeleton data, not by construction from the inputs.

full rationale

This is an empirical supervised-learning paper, not a derivation chain. The Transformer is trained on synchronized foot pressure, IMU, and OptiTrack skeleton data, and accuracy is measured on a held-out portion of the data (Section V-D: 'The dataset was split into training and validation sets in an 8:2 ratio'). There is no step in which a predicted quantity is defined in terms of itself, no fitted parameter is renamed as a prediction, and no uniqueness theorem is imported from the authors' prior work. The only self-citation is reference [23], PIFall, by Guo, Liu, Lu, and Jing; it appears in related work as an example of pressure-based fall detection and is not load-bearing for the Transformer architecture, the derivative-augmented input, or the evaluation protocol. The claims about first- and second-order derivatives are supported by a controlled comparison (Figure 8), and the robustness claims are supported by task-wise RMSE tables, which are empirical results rather than identities. The concern that the random 8:2 frame-level split may allow temporal leakage from the same participants and movement sequences is a generalization and validity concern, not circularity: even if the reported RMSE overstates performance on new people, the evaluation is not equivalent to the training objective by construction. The conclusion explicitly acknowledges that 'expanding the dataset regarding the diversity of motions and participants would improve robustness and generalization,' confirming the limitation is recognized. Therefore, no specific circular step can be quoted, and the appropriate score is 0.

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

This is an applied ML paper, so the main axioms are domain assumptions about sensor information content and data representativeness, plus preprocessing choices. The free parameters are the manually chosen architecture and training hyperparameters, the data split, and the learned network weights; none are derived from theory or released for independent checking.

free parameters (4)
  • Transformer architecture hyperparameters = d_model=512, encoder_layers=8, heads=8, dropout=0.1, feedforward=4x d_model
    Chosen from established practice and preliminary experiments (Section IV); no sensitivity analysis is provided, so the reported accuracy may depend on these choices.
  • Optimizer and training hyperparameters = lr=0.0005, weight_decay=0.001, epochs=200, batch_size=32, AdamW, ReduceLROnPlateau
    Fixed without reported tuning curves; the final model is selected on validation loss (Section IV), so these choices directly affect the reported RMSE.
  • Data splitting ratio = 8:2 train/validation
    Random split over frames from four participants; no subject-independent split, and this choice is the main source of optimistic error estimates (Section V-D).
  • Neural network weights = Not released; learned via AdamW on the training split
    The central mapping from 82-dim input to 63-dim output is entirely learned; since no pretrained weights or architecture details beyond the text are provided, the result is inseparable from this specific training run.
assumptions (5)
  • domain assumption Foot plantar pressure and ankle IMU signals contain sufficient information to determine full-body 3D pose (including head and arms).
    This is the core feasibility premise for the whole approach; it is not derived or validated with a kinematic analysis and is only partially supported by the reported RMSE values (Section V).
  • domain assumption The OptiTrack labels and insole/IMU streams are accurately synchronized and co-registered.
    The pipeline synchronizes by timestamp and downsampling (Section V-C), but no synchronization error analysis is provided; label misalignment would directly bias the reported errors.
  • ad hoc to paper Four participants performing eight movements for 20 minutes each are representative of the target daily-life population.
    The dataset scale and participant diversity are stated in Section V-B without justification; the system is claimed for rehabilitation and health monitoring, but no cross-subject generalization test is run.
  • domain assumption The 35 pressure sensors cover foot deformation relevant to all tested postures.
    Sensor placement is fixed (Section III), and the paper mentions optimizing placement but shows no analysis; the contribution bullet about sensor deployment and error correlation is not present in the experiment section.
  • domain assumption The non-inverting amplifier circuit and Velostat sensors respond linearly over the pressure range.
    Equation (1) gives the circuit relation, but no calibration curve or linearity test is presented (Section III).

how reviews work

0 comments
Cite this review

Pith. "Pith review of P2P-Insole: Human Pose Estimation Using Foot Pressure Distribution and Motion Sensors." pith.science (2026). https://pith.science/paper/E2FZBGSN

@misc{pith2026250500755,
  author       = {Pith},
  title        = {Pith review of: P2P-Insole: Human Pose Estimation Using Foot Pressure Distribution and Motion Sensors},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/E2FZBGSN}},
  note         = {Machine review of arXiv:2505.00755}
}
read the original abstract

This work presents P2P-Insole, a low-cost approach for estimating and visualizing 3D human skeletal data using insole-type sensors integrated with IMUs. Each insole, fabricated with e-textile garment techniques, costs under USD 1, making it significantly cheaper than commercial alternatives and ideal for large-scale production. Our approach uses foot pressure distribution, acceleration, and rotation data to overcome limitations, providing a lightweight, minimally intrusive, and privacy-aware solution. The system employs a Transformer model for efficient temporal feature extraction, enriched by first and second derivatives in the input stream. Including multimodal information, such as accelerometers and rotational measurements, improves the accuracy of complex motion pattern recognition. These facts are demonstrated experimentally, while error metrics show the robustness of the approach in various posture estimation tasks. This work could be the foundation for a low-cost, practical application in rehabilitation, injury prevention, and health monitoring while enabling further development through sensor optimization and expanded datasets.

Figures

Figures reproduced from arXiv: 2505.00755 by the authors.

Figure 1
Figure 1. shows the system’s architecture of the P2P-Insole. The system inputs foot pressure distribution, acceleration, and rotation information collected by an insole sensor and an inertial measurement unit (IMU). It uses a Transformer model to predict a 3D skeleton. The model is trained using a dataset that simultaneously captures data collected from the insole sensor and IMU, along with the corresponding 3D skeleton data … view at source ↗
Figure 2
Figure 2. Insole Sensor Design and Non-Inverting Amplifier Circuit [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Experimental Setup The training process utilized Mean Squared Error (MSE) as the loss function, optimized with the AdamW optimizer. The learning rate was initialized at 0.0005, with a weight decay of 0.001. Additionally, a ReduceLROnPlateau scheduler dynamically adjusted the learning rate when no improvement in validation loss was observed. The model was trained for 200 epochs with a batch size of 32. The dataset wa… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Transformer Architecture Red: Predict skeleton Green: True skeleton Pressure Task Stand Tilt Bow Stand and Sit Squat Blue: Acceleration Red: Gyro Left Right Left Right Left Right Left Right Left Right [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Highly accurate instantaneous predicted 3D skeleton. [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 6
Figure 6. Figure 6: Pose Label Generation then downsampled to match the 0.01-second intervals used for the 3D skeleton data. To facilitate efficient model training, normalization (eq. 3) and standardization (eq. 4) were applied across all data points. Additionally, feature enhancement tec…
Figure 7
Figure 7. Figure 7: RMSE Comparison of Transformer and LSTM -10 10 30 50 70 90 110 130 150 Stand Tilt Bow Stand and Sit Squat Average With Derivative No Derivative (mm) [PITH_FULL_IMAGE:figures/full_fig_p005_7.png]
Figure 8
Figure 8. Figure 8: RMSE of comparison with and without derivative data [PITH_FULL_IMAGE:figures/full_fig_p005_8.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Towards Real-World Wearable Motion Reconstruction

    cs.CV 2026-07 conditional novelty 6.0 of 10

    A consumer-wearable MoCap dataset plus WHIP, a flow-matching model that reconstructs full-body motion from arbitrary sensor subsets and quantifies sensor complementarity.

Reference graph

Works this paper leans on

24 extracted references · 23 canonical work pages · cited by 1 Pith paper

  1. [6]

    Soleposer: Full body pose estimation using a single pair of insole sensor,

    E. Wu, R. Khirodkar, H. Koike, and K. Kitani, “Soleposer: Full body pose estimation using a single pair of insole sensor,” in Proceedings of the 37th Annual ACM Symposium on User Interface Software and Technology, pp. 1–9, 2024

    work page 2024

  2. [1]

    Realtime multi-person 2d pose estimation using part affinity fields,

    Z. Cao, T. Simon, S.-E. Wei, and Y . Sheikh, “Realtime multi-person 2d pose estimation using part affinity fields,” in Proceedings of the IEEE conference on computer vision and pattern recognition , pp. 7291–7299, 2017

    2017

  3. [2]

    Learnable trian- gulation of human pose,

    K. Iskakov, E. Burkov, V . Lempitsky, and Y . Malkov, “Learnable trian- gulation of human pose,” in Proceedings of the IEEE/CVF international conference on computer vision , pp. 7718–7727, 2019

    work page 2019

  4. [3]

    Tessetrack: End-to-end learnable multi-person articulated 3d pose tracking,

    N. D. Reddy, L. Guigues, L. Pishchulin, J. Eledath, and S. G. Narasimhan, “Tessetrack: End-to-end learnable multi-person articulated 3d pose tracking,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. 15190–15200, 2021

    work page 2021

  5. [4]

    Mmvp: A multimodal mocap dataset with vision and pressure sensors,

    H. Zhang, S. Ren, H. Yuan, J. Zhao, F. Li, S. Sun, Z. Liang, T. Yu, Q. Shen, and X. Cao, “Mmvp: A multimodal mocap dataset with vision and pressure sensors,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. 21842–21852, 2024

    work page 2024

  6. [5]

    Identity-preserved human posture detection in infrared thermal images: A benchmark,

    Y . Guo, Y . Chen, J. Deng, S. Li, and H. Zhou, “Identity-preserved human posture detection in infrared thermal images: A benchmark,” Sensors, vol. 23, no. 1, p. 92, 2022

    work page 2022

  7. [7]

    Deep inertial poser: Learning to reconstruct human pose from sparse inertial measurements in real time,

    Y . Huang, M. Kaufmann, E. Aksan, M. J. Black, O. Hilliges, and G. Pons-Moll, “Deep inertial poser: Learning to reconstruct human pose from sparse inertial measurements in real time,” ACM Transactions on Graphics (TOG), vol. 37, no. 6, pp. 1–15, 2018

    work page 2018

  8. [8]

    Hperl: 3d human pose estimation from rgb and lidar,

    M. F ¨urst, S. T. Gupta, R. Schuster, O. Wasenm ¨uller, and D. Stricker, “Hperl: 3d human pose estimation from rgb and lidar,” in 2020 25th International Conference on Pattern Recognition (ICPR) , pp. 7321– 7327, IEEE, 2021

    work page 2020

Show all 24 references
  1. [9]

    Fusing monocular images and sparse imu signals for real-time human motion capture,

    S. Pan, Q. Ma, X. Yi, W. Hu, X. Wang, X. Zhou, J. Li, and F. Xu, “Fusing monocular images and sparse imu signals for real-time human motion capture,” in SIGGRAPH Asia 2023 Conference Papers, pp. 1–11, 2023

    2023

  2. [10]

    Populating 3d scenes by learning human-scene interaction,

    M. Hassan, P. Ghosh, J. Tesch, D. Tzionas, and M. J. Black, “Populating 3d scenes by learning human-scene interaction,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. 14708–14718, 2021

    2021

  3. [11]

    In-bed pressure-based pose estimation using image space representation learning,

    V . Davoodnia, S. Ghorbani, and A. Etemad, “In-bed pressure-based pose estimation using image space representation learning,” in ICASSP 2021- 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 3965–3969, IEEE, 2021

    2021

  4. [12]

    Pressure-based in-bed pose estimation methods using the skinned multi-person linear model,

    C. Lei, F. Li, X. Chen, and L. Yang, “Pressure-based in-bed pose estimation methods using the skinned multi-person linear model,” in 2024 4th Asia Conference on Information Engineering (ACIE) , pp. 74– 79, IEEE, 2024

    2024

  5. [13]

    3d human pose estimation using pressure images on a smart chair,

    M. Zhao, F. Xie, Z. Wu, Z. Liang, and X. Cai, “3d human pose estimation using pressure images on a smart chair,” in Proceedings of the 2024 2nd Asia Conference on Computer Vision, Image Processing and Pattern Recognition , pp. 1–7, 2024

    2024

  6. [14]

    Intelligent seat: Tactile signal-based 3d sitting pose inference,

    M. Seong, G. Kim, J. Lee, J. DelPreto, W. Matusik, D. Rus, and S. Kim, “Intelligent seat: Tactile signal-based 3d sitting pose inference,” in Companion of the 2024 on ACM International Joint Conference on Pervasive and Ubiquitous Computing , pp. 791–796, 2024

    2024

  7. [15]

    Machine vision-based human action recognition using spatio- temporal motion features (stmf) with difference intensity distance group pattern (didgp),

    J. Arunnehru, S. Thalapathiraj, R. Dhanasekar, L. Vijayaraja, R. Kan- nadasan, A. A. Khan, M. A. Haq, M. Alshehri, M. I. Alwanain, and I. Keshta, “Machine vision-based human action recognition using spatio- temporal motion features (stmf) with difference intensity distance gro...

    2022

  8. [16]

    3d human pose estimation with spatio-temporal criss-cross attention,

    Z. Tang, Z. Qiu, Y . Hao, R. Hong, and T. Yao, “3d human pose estimation with spatio-temporal criss-cross attention,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. 4790–4799, 2023

    2023

  9. [17]

    Faster voxelpose: Real- time 3d human pose estimation by orthographic projection,

    H. Ye, W. Zhu, C. Wang, R. Wu, and Y . Wang, “Faster voxelpose: Real- time 3d human pose estimation by orthographic projection,” in European Conference on Computer Vision , pp. 142–159, Springer, 2022

    2022

  10. [18]

    Fusion of multiple lidars and inertial sensors for the real-time pose tracking of human motion,

    A. K. Patil, A. Balasubramanyam, J. Y . Ryu, P. K. BN, B. Chakravarthi, and Y . H. Chai, “Fusion of multiple lidars and inertial sensors for the real-time pose tracking of human motion,” Sensors, vol. 20, no. 18, p. 5342, 2020

    2020

  11. [19]

    Satpose: Improving monocular 3d pose estimation with spatial-aware ground tactility,

    L. Zhan, E. Ying, J. Gan, S. Guo, B. Gao, and Y . Qin, “Satpose: Improving monocular 3d pose estimation with spatial-aware ground tactility,” in Proceedings of the 32nd ACM International Conference on Multimedia , pp. 6192–6201, 2024

    2024

  12. [20]

    Intelligent carpet: Inferring 3d human pose from tactile signals,

    Y . Luo, Y . Li, M. Foshey, W. Shou, P. Sharma, T. Palacios, A. Torralba, and W. Matusik, “Intelligent carpet: Inferring 3d human pose from tactile signals,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. 11255–11265, 2021

    2021

  13. [21]

    Human pose estimation from ambigu- ous pressure recordings with spatio-temporal masked transformers,

    V . Davoodnia and A. Etemad, “Human pose estimation from ambigu- ous pressure recordings with spatio-temporal masked transformers,” in ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pp. 1–5, IEEE, 2023

    2023

  14. [22]

    Dual-modal 3d human pose estimation using insole foot pressure sensors,

    E. Wu, Y . Peng, R. Khirodkar, H. Koike, and K. Kitani, “Dual-modal 3d human pose estimation using insole foot pressure sensors,” in 2024 IEEE International Symposium on Mixed and Augmented Reality Adjunct (ISMAR-Adjunct), pp. 131–135, IEEE, 2024

    2024

  15. [23]

    Pifall: A pressure insole-based fall detection system for the elderly using resnet3d,

    W. Guo, X. Liu, C. Lu, and L. Jing, “Pifall: A pressure insole-based fall detection system for the elderly using resnet3d,” Electronics, vol. 13, no. 6, p. 1066, 2024

    2024

  16. [24]

    Learn to infer human poses using a full-body pressure sensing garment,

    D. Zhang, Z. Liang, Y . Wu, F. Xie, G. Xu, Z. Wu, and X. Cai, “Learn to infer human poses using a full-body pressure sensing garment,” IEEE Sensors Journal, 2024. Atsuya Watanabe received his B.S. degree in Com- puter Science and Engineering from the University of Aizu, Japan,...

    2024

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.