Pith. sign in

REVIEW 4 major objections 4 minor 50 references

PoseGRAF: Geometric-Reinforced Adaptive Fusion for Monocular 3D Human Pose Estimation

T0 review · 4 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read PoseGRAF claims that adding bone-direction and joint-angle graphs to a 2D-to-3D lifting network improves monocular 3D pose accuracy, reporting top MPJPE, P-MPJPE, and PCK results on Human3.6M and MPI-INF-3DHP.

desk verdict Competent incremental pose lifting paper with plausible small gains; the cross-dataset SOTA claim is under-specified and the GraFormer comparison is misquoted. read the letter →

arxiv 2506.14596 v1 pith:6LYCMLVY submitted 2025-06-17 cs.CV cs.AI

Ming Xu , Xu Zhang This is my paper · ORCID
classification cs.CVcs.AI
keywords 3Dhumanposeestimationmonocular2D-to-3Dliftinggraphconvolutionalnetworksbonedirectionjointanglesdynamicfeaturefusioncross-attentionHuman3.6M
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 tries to establish that monocular 3D human pose estimation from 2D keypoints improves when the skeleton is represented not only by joint positions but also by bone directions and the angles between adjacent bones. It builds a dual-graph architecture, a cross-attention mechanism between joint and bone features, and an adaptive fusion stage that selects key joints and reconstructs global features by traversing the skeleton graph. On Human3.6M, the method reports the best MPJPE and P-MPJPE with CPN-detected 2D poses (48.1 mm and 38.3 mm), and on MPI-INF-3DHP it reports the best PCK (88.3) among the compared methods. The claim matters because implausible poses under occlusion and fast motion are a known failure mode of lifting methods, and the paper's design targets that failure directly.

What carries the argument

The load-bearing component is a Bone Direction GCN that builds two bone graphs: a directed weighted graph whose edge weights are the angles between adjacent bone direction vectors, computed via $\arccos$ of their normalized inner product, and a directed unweighted graph of bone connectivity. Beside it, a Joint GCN processes the usual joint-position graph. A Cross-Attention module mixes joint and bone features, and a Dynamic Fusion module selects the top-$\mu$ joints by attention score, then propagates from those seeds across the joint graph with a breadth-first search while injecting bone-direction features, and adds the original joint features back. An improved Transformer encoder rescales attention weights with a relative distance matrix derived from the skeleton topology and is joined to the rest in a residual manner.

What would settle it

On Human3.6M ground-truth 2D input, sweep the unspecified parameters—$w$ from 0 to 5 and $\mu$ from 1 to 17—and re-run the MPI-INF-3DHP transfer with the Human3.6M joint ordering explicitly remapped; if MPJPE or PCK moves by more than the reported margins over GraphMLP, the headline gains are not robust to these undisclosed choices.

Watch

Extended reading notes

Core claim

The central claim, stated on the paper's own terms, is that explicitly modeling bone directions and the angles between adjacent bones as weighted edges in a graph, then adaptively fusing those geometric features with joint-position features through cross-attention and a top-$\mu$ breadth-first reconstruction, lets a 2D-to-3D lifting network exceed the prior state of the art on both Human3.6M and MPI-INF-3DHP. The reported numbers are 48.1 mm MPJPE and 38.3 mm P-MPJPE with CPN-detected 2D poses on Human3.6M, 32.1 mm MPJPE with ground-truth 2D poses, and 88.3 PCK with 54.8 AUC on MPI-INF-3DHP using a model trained on Human3.6M with no stated adaptation.

Load-bearing premise

The result rests on the unstated assumptions that the two datasets share an identical joint set, ordering, and skeletal connectivity used to build the joint graph, bone graphs, and BFS, and that the distance-scaling weight $w$ and top-$\mu$ selection are set appropriately.

Editorial extensions

If this is right

  • With CPN-detected 2D input on Human3.6M, PoseGRAF reports 48.1 mm MPJPE and 38.3 mm P-MPJPE, the best averages in the comparison tables.
  • With ground-truth 2D input, it reports 32.1 mm MPJPE and 25.0 mm P-MPJPE, a 7.2% reduction over its own Transformer-plus-Joint-GCN baseline.
  • A Human3.6M-trained model reports 88.3% PCK and 54.8% AUC on MPI-INF-3DHP with no stated adaptation to the second dataset.
  • Ablations show that adding the Bone GCN and Cross-Attention lowers error monotonically and that the full dynamic-fusion configuration is the best performer, supporting the design's core premise.

Reading between the lines

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

  • A testable extension the paper does not run: sweep the unspecified parameters $w$ and $\mu$ and report the error surface; if the gains over GraphMLP vanish outside a narrow band, the adaptive fusion is not the true source of improvement.
  • Because every module assumes a fixed skeleton graph, transferring to a dataset with a different joint convention would require remapping joint indices and recomputing the bone graphs; the paper does not discuss this implicit cost.
  • The comparisons are single-frame; inserting the same geometric fusion into a temporal encoder, as other lifted pose models do, is a natural next step that the paper leaves open.
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 / 4 minor

Summary. This paper proposes PoseGRAF, a 2D-to-3D monocular human pose estimation framework that combines joint and bone graph convolutions, a cross-attention module, an attention-based dynamic fusion module, and a transformer encoder with relative-distance attention scaling. The authors report state-of-the-art MPJPE (48.1 mm) and P-MPJPE (38.3 mm) on Human3.6M with CPN-detected 2D poses, and PCK (88.3%) and AUC (54.8%) on MPI-INF-3DHP when a Human3.6M-trained model is applied directly. Ablations on Human3.6M ground-truth 2D poses attribute gains to the bone-direction GCN, cross-attention, and dynamic fusion. The code is stated to be publicly available.

Significance. If the reported numbers are reproducible, PoseGRAF offers a modest but useful improvement over the closest competitor GraphMLP on both benchmarks, and the ablation study provides evidence that the geometric bone-direction modeling contributes positively. The release of code is a practical strength. However, the margins over prior work are small (1.1 mm MPJPE, 0.3 mm P-MPJPE, 1.3 PCK, 0.5 AUC), the paper overstates the GraFormer comparison, and key hyperparameters and the cross-dataset protocol are underspecified. The central architectural idea is defensible, but the current presentation does not yet provide enough detail to verify the strongest claims.

major comments (4)
  1. [4.3, Table 1] The claim in Section 4.3 that PoseGRAF achieves 'a reduction of 10.6 mm in MPJPE over GraFormer' is arithmetically inconsistent with Table 1, which lists GraFormer at 51.8 mm and PoseGRAF at 48.1 mm, a gap of 3.7 mm. The text should either be corrected to 3.7 mm or the intended comparison should be stated explicitly.
  2. [4.3, Table 3] The MPI-INF-3DHP evaluation is described as applying the Human3.6M-trained model 'directly' (Section 4.3), but the manuscript does not specify the joint correspondence, the joint count N, the skeletal adjacency matrix G_J, or the bone-node construction used for MPI-INF-3DHP. Since Eqs. (1)-(10) and Algorithm 1 depend on G_J and the bone graph, the reported 88.3 PCK is not verifiable against prior comparisons unless these mapping details are provided.
  3. [3.4, Algorithm 1 and 3.5, Eq. (11)] The hyperparameters w in Eq. (11) and the top-μ selection in Algorithm 1 are load-bearing but never specified. Section 3.5 states only that 'we set w to facilitate information exchange...' without a value or tuning procedure, and Section 3.4 describes μ as both a 'learnable threshold parameter' and the count of selected top-μ joint features, while Algorithm 1's Top_Indices call does not define μ. Please provide explicit values, ranges, or learning rules for both.
  4. [Tables 1-3] The reported improvements over GraphMLP are small (1.1 mm MPJPE, 0.3 mm P-MPJPE, 1.3 PCK, 0.5 AUC), yet no error bars, multiple-seed variation, or significance tests are reported. Without such information, the claim of 'exceeding state-of-the-art' is not robust to run-to-run variability; at minimum, the authors should report standard deviations across seeds for the main tables.
minor comments (4)
  1. [Eq. (3)] The denominator of the angle formula should be ‖x_B^p‖‖x_B^q‖, not ‖x_B^p‖‖x_B^p‖.
  2. [Tables 1-2] The caption of Table 1 uses '&' and '*' markers, while Table 2's caption uses '*'; ensure the symbols are defined consistently across tables and that the rows actually carry the markers.
  3. [Section 4.4, Table 4] The text says static fusion 'increases MPJPE by 4.1 mm' compared to dynamic fusion, but Table 4 does not contain the static-fusion pair needed to verify this number; please add the missing ablation row or clarify the reference point.
  4. [Section 4.2] Implementation details list architecture and training hyperparameters but omit the values of w and μ; these should be included alongside the other settings.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: PoseGRAF's benchmark claims rest on standard external train/test evaluation and independently cited baselines, not on self-referential reductions.

full rationale

I walked the paper's derivation chain: 2D CPN keypoints are converted into joint and bone graphs (Eqs. 2-3), processed by dual GCNs (Eq. 4), Cross-Attention (Eqs. 5-8), Dynamic Fusion (Eqs. 9-10 and Algorithm 1), a Transformer with a distance-aware rescaling (Eqs. 11-13), and a standard MPJPE loss (Eq. 14). No equation defines a claimed prediction in terms of the training labels, and no fitted parameter is later renamed as a prediction. The Human3.6M and MPI-INF-3DHP results in Tables 1-3 are standard held-out benchmark numbers compared against published state-of-the-art results; the ablation in Table 4 compares architectural variants under the same protocol. The under-specified hyperparameters w (Eq. 11) and the top-mu selection (Algorithm 1), together with the missing explicit joint mapping for the MPI-INF-3DHP cross-dataset test, are reproducibility and validity concerns rather than circularity. Citations such as [31], [32], and [34] motivate architectural components but do not carry the empirical claim; there is no self-citation chain that forces the result. The central claim is therefore self-contained against external benchmarks, and no circular step can be exhibited from the paper's own equations or citations.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claim relies on the transferability of a fixed skeleton graph across datasets, on the informativeness of 2D bone angles for 3D pose, and on the BFS-based reconstruction heuristic. Two hyperparameters (w and mu) are load-bearing but not specified. These are modeling choices, not independent evidence.

free parameters (2)
  • w = not specified
    Eq. (11) introduces the distance-aware intensity hyperparameter w that rescales the joint adjacency matrix A_J inside the Transformer attention. The text says 'we set w to facilitate information exchange between non-local nodes' but never gives a value, range, or tuning procedure. The reported accuracy likely depends on this choice.
  • mu (top-mu selection) = not specified
    Algorithm 1 and Section 3.4 describe a 'learnable threshold parameter mu' that selects the top-mu joint features for the dynamic fusion. The paper does not state whether mu is a count or a score threshold, its initial value, or how it is optimized. The ablation in Table 4 credits this selection with a meaningful part of the gain.
assumptions (3)
  • domain assumption The human skeleton adjacency A_J is a fixed prior shared by the training and test datasets
    Section 3.2 defines the joint graph; Section 4.3 directly applies a Human3.6M-trained model to MPI-INF-3DHP, which is only valid if both datasets use the same joint set, ordering, and connectivity.
  • domain assumption 2D bone directions and inter-bone angles are informative features for 3D pose prediction
    The bone-graph branch encodes normalized 2D bone vectors (Eq. 2) and inter-bone angles (Eq. 3) as inputs; the entire method rests on this geometric signal transferring from 2D to 3D.
  • domain assumption Breadth-first traversal along the skeleton with bone features reconstructs a useful global joint descriptor
    Eq. (9)-(10) and Algorithm 1 assume that BFS seeded at selected joints and guided by the fixed skeleton graph G_J, while incorporating bone features X_B, produces a feature that improves regression. No evidence is given that this particular graph traversal is optimal or even beneficial beyond the ablation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PoseGRAF: Geometric-Reinforced Adaptive Fusion for Monocular 3D Human Pose Estimation." pith.science (2026). https://pith.science/paper/6LYCMLVY

@misc{pith2026250614596,
  author       = {Pith},
  title        = {Pith review of: PoseGRAF: Geometric-Reinforced Adaptive Fusion for Monocular 3D Human Pose Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6LYCMLVY}},
  note         = {Machine review of arXiv:2506.14596}
}
read the original abstract

Existing monocular 3D pose estimation methods primarily rely on joint positional features, while overlooking intrinsic directional and angular correlations within the skeleton. As a result, they often produce implausible poses under joint occlusions or rapid motion changes. To address these challenges, we propose the PoseGRAF framework. We first construct a dual graph convolutional structure that separately processes joint and bone graphs, effectively capturing their local dependencies. A Cross-Attention module is then introduced to model interdependencies between bone directions and joint features. Building upon this, a dynamic fusion module is designed to adaptively integrate both feature types by leveraging the relational dependencies between joints and bones. An improved Transformer encoder is further incorporated in a residual manner to generate the final output. Experimental results on the Human3.6M and MPI-INF-3DHP datasets show that our method exceeds state-of-the-art approaches. Additional evaluations on in-the-wild videos further validate its generalizability. The code is publicly available at https://github.com/iCityLab/PoseGRAF.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 48 canonical work pages

  1. [1]

    Du, S., Yuan, Z., & Ikenaga, T. (2024). *Kinematics-aware spatial -temporal feature transform for 3D human pose estimation*. Pattern Recognition, 150, 110316

    work page 2024

  2. [2]

    𝐼𝑛𝑑𝑒𝑥 = 𝑇𝑜𝑝_𝐼𝑛𝑑𝑖𝑐𝑒𝑠(𝑆ℎ𝑖𝑔ℎ, 𝑋𝐽𝐶)

    work page

  3. [3]

    𝒳𝐽𝐶 = {𝑋𝐽𝐶 𝑖 |𝑖 ∈ 𝐼𝑛𝑑𝑒𝑥}

    work page

  4. [4]

    Experiments 4.1 Datasets and evaluation metrics This section presents comprehensive studies on two real -world 3D human pose estimation benchmark datasets to systematically validate the superiority of the proposed model. Human3.6M Datasets: As the most representative benchmark in 3D human pose estimation, the Human3.6M Dataset [13] provides 3.6 million fr...

    work page

  5. [5]

    Xu, X., Liu, L., & Yan, S. (2023). SMPlER: Taming Transformers for Monocular 3D Human Shape and Pose Estimation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(5), 3275–3289

    work page 2023

  6. [6]

    𝒥𝐵 𝑖 = 𝐵𝐹𝑆(𝐺𝐽, 𝑋𝐽𝐶 𝑖 , 𝑋𝐵)

    work page

  7. [7]

    M., Nooruddin, S., Karray, F., & Muhammad, G

    Islam, M. M., Nooruddin, S., Karray, F., & Muhammad, G. (2023). Multi-level feature fusion for multimodal human activity recognition in Internet of Healthcare Things . Information Fusion, 94, 17–31

    work page 2023

  8. [8]

    Xu, C., He, J., Zhang, X., Yao, C., & Tseng, P. H. (2018). Geometrical kinematic modeling on human motion using method of multi-sensor fusion. Information Fusion, 41, 243–254

    work page 2018

Show all 50 references
  1. [9]

    𝑋𝐷𝐹=∑ 𝒥𝐵 𝑖𝜇−1 𝑖=0 +𝑋𝐽𝐶 3.5 Transformer Encoder The conventional Transformer encoder can model global dependencies through multi -head self - attention, enabling each node to equally influence others. However, its permutation -invariant property neglects the critical topologica...

  2. [10]

    P ., Muthukrishnan, S

    Tripathi, A., Prathosh, A. P ., Muthukrishnan, S. P., & Kumar, L. (2023). SurfMyoAiR: A surface electromyography-based framework for airwriting recognition . IEEE Transactions on Instrumentation and Measurement, 72, 1–12

    2023

  3. [11]

    Y ., & Lee, K

    Moon, G., Chang, J. Y ., & Lee, K. M. (2019). *Camera Distance-Aware Top-Down Approach for 3D Multi -Person Pose Estimation from a Single RGB Image*. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 10133–10142)

    2019

  4. [12]

    Zeng, A., Sun, X., Huang, F., Liu, M., Xu, Q., & Lin, S. (2020). *SRNet: Improving generalization in 3D human pose estimation with a split -and-recombine approach*. In Computer Vision – ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XIV...

    2020

  5. [13]

    Chen, Z., Dai, J., Bai, J., & Pan, J. (2024). DGFormer: Dynamic graph transformer for 3D human pose estimation. Pattern Recognition, 152, 110446

    2024

  6. [14]

    Zheng, C., Zhu, S., Mendieta, M., Yang, T., Chen, C., & Ding, Z. (2021). 3D Human Pose Estimation with Spatial and Temporal Transformers . In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 11656–11665)

    2021

  7. [15]

    Sun, X., Shang, J., Liang, S., & Wei, Y . (2017). Compositional Human Pose Regression . In Proceedings of the IEEE International Conference on Computer Vision (pp. 2602–2611)

    2017

  8. [16]

    Chen, T., Fang, C., Shen, X., Zhu, Y ., Chen, Z., & Luo, J. (2021). *Anatomy-aware 3D human pose estimation with bone -based pose decomposition*. IEEE Transactions on Circuits and Systems for Video Technology, 32(1), 198–209

    2021

  9. [17]

    Xu, J., Yu, Z., Ni, B., Yang, J., Yang, X., & Zhang, W. (2020). Deep kinematics analysis for monocular 3D human pose estimation . In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 899–908)

    2020

  10. [18]

    Ionescu, C., Papava, D., Olaru, V ., & Sminchisescu, C. (2013). Human3.6M: Large Scale Datasets and Predictive Methods for 3D Human Sensing in Natural Environments . IEEE Transactions on Pattern Analysis and Machine Intelligence, 36(7), 1325–1339

    2013

  11. [19]

    Mehta, D., Rhodin, H., Casas, D., Fua, P., Sotnychenko, O., Xu, W., & Theobalt, C. (2017). Monocular 3D Human Pose Estimation in the Wild Using Improved CNN Supervision. In Proceedings of the International Conference on 3D Vision (3DV) (pp. 506– 516)

    2017

  12. [20]

    Xu, X., & Loy, C. C. (2021). 3D human texture estimation from a single image with transformers. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 13849–13858)

    2021

  13. [21]

    H., & Ramanan, D

    Chen, C. H., & Ramanan, D. (2017). *3D Human Pose Estimation = 2D Pose Estimation + Matching*. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 7035–7043)

    2017

  14. [22]

    Martinez, J., Hossain, R., Romero, J., & Little, J. J. (2017). A simple yet effective baseline for 3D human pose estimation . In Proceedings of the IEEE International Conference on Computer Vision (pp. 2640–2649)

    2017

  15. [23]

    Lee, K., Lee, I., & Lee, S. (2018). Propagating LSTM: 3D Pose Estimation Based on Joint Interdependency. In Proceedings of the European Conference on Computer Vision (ECCV) (pp. 119–135)

    2018

  16. [24]

    Wei, G., Lan, C., Zeng, W., & Chen, Z. (2019). View Invariant 3D Human Pose Estimation. IEEE Transactions on Circuits and Systems for Video Technology, 30(12), 4601–4610

    2019

  17. [25]

    Gong, K., Zhang, J., & Feng, J. (2021). PoseAug: A Differentiable Pose Augmentation Framework for 3D Human Pose Estimation . In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 8575–8584)

    2021

  18. [26]

    Ci, H., Wang, C., Ma, X., & Wang, Y . (2019). Optimizing Network Structure for 3D Human Pose Estimation. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 2262–2271)

    2019

  19. [27]

    Zhao, L., Peng, X., Tian, Y ., Kapadia, M., & Metaxas, D. N. (2019). Semantic Graph Convolutional Networks for 3D Human Pose Regression. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 3425–3435)

    2019

  20. [28]

    Zou, Z., & Tang, W. (2021). Modulated Graph Convolutional Network for 3D Human Pose Estimation. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 11477–11487). https://doi.org/10.1109/ICCV48922.2021.01128

    2021

  21. [29]

    Q., Li, C., Fang, Y ., Geng, Y ., Xie, X., & Keuper, M

    Zhou, Y ., Cheng, Z. Q., Li, C., Fang, Y ., Geng, Y ., Xie, X., & Keuper, M. (2022). Hypergraph Transformer for Skeleton-Based Action Recognition. arXiv preprint arXiv:2211.09590

    2022 arXiv

  22. [30]

    Ma, X., Su, J., Wang, C., Ci, H., & Wang, Y . (2021). Context Modeling in 3D Human Pose Estimation: A Unified Perspective . In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 6238–6247)

    2021

  23. [31]

    Azizi, N., Possegger, H., Rodolà, E., & Bischof, H. (2022). 3D Human Pose Estimation Using Möbius Graph Convolutional Networks. In Proceedings of the European Conference on Computer Vision (ECCV) (pp. 160–178)

    2022

  24. [32]

    Hu, W., Zhang, C., Zhan, F., Zhang, L., & Wong, T. T. (2021). Conditional Directed Graph Convolution for 3D Human Pose Estimation. In Proceedings of the 29th ACM International Conference on Multimedia (pp. 602–611)

    2021

  25. [33]

    X., Zhang, Z., Liu, Y ., Zhong, S

    Yu, B. X., Zhang, Z., Liu, Y ., Zhong, S. H., Liu, Y ., & Chen, C. W. (2023). *GLA-GCN: Global-Local Adaptive Graph Convolutional Network for 3D Human Pose Estimation from Monocular Video*. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 8818–8829)

    2023

  26. [34]

    J., Jacobs, D

    Kanazawa, A., Black, M. J., Jacobs, D. W., & Malik, J. (2018). End-to-End Recovery of Human Shape and Pose . In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 7122–7131)

    2018

  27. [35]

    Chen, Y ., Wang, Z., Peng, Y ., Zhang, Z., Yu, G., & Sun, J. (2018). Cascaded Pyramid Network for Multi -Person Pose Estimation . In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 7103–7112)

    2018

  28. [36]

    Chen, M., Lin, M., Li, K., Shen, Y ., Wu, Y ., Chao, F., & Ji, R. (2023). CF-ViT: A General Coarse-to-Fine Method for Vision Transformer. In Proceedings of the AAAI Conference on Artificial Intelligence, 37(6), 7042–7052

    2023

  29. [37]

    An, X., Zhao, L., Gong, C., Wang, N., Wang, D., & Yang, J. (2024). Sharpose: Sparse high- resolution representation for human pose estimation. In Proceedings of the AAAI Conference on Artificial Intelligence, 38(2), 691–699

    2024

  30. [38]

    N., Kaiser, L., & Polosukhin, I

    Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017). Attention Is All You Need. Advances in Neural Information Processing Systems, 30

    2017

  31. [39]

    Zhang, Z., et al. (2021). Motif-based graph self-supervised learning for molecular property prediction. Advances in Neural Information Processing Systems, 34, 15870–15882

    2021

  32. [40]

    Xu, T., & Takano, W. (2021). Graph Stacked Hourglass Networks for 3D Human Pose Estimation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 16105–16114)

    2021

  33. [41]

    Zhao, W., Wang, W., & Tian, Y . (2022). *Graformer: Graph-Oriented Transformer for 3D Pose Estimation*. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 20438–20447)

    2022

  34. [42]

    Zhang, J., Tu, Z., Yang, J., Chen, Y ., & Yuan, J. (2022). *MixSTE: Seq2seq Mixed Spatio- Temporal Encoder for 3D Human Pose Estimation in Video*. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 13232–13242)

    2022

  35. [43]

    Li, H., Shi, B., Dai, W., Zheng, H., Wang, B., Guo, M., Li, C., Zou, J., & Xiong, H. (2023). *Pose-Oriented Transformer with Uncertainty -Guided Refinement for 2D -to-3D Human Pose Estimation*. In Proceedings of the AAAI Conference on Artificial Intelligence, 37(1), 1296–1304

    2023

  36. [44]

    T., & Hamza, A

    Hassan, M. T., & Hamza, A. B. (2023). Regular Splitting Graph Network for 3D Human Pose Estimation. IEEE Transactions on Image Processing, 32, 4212–4222

    2023

  37. [45]

    Li, W., Liu, M., Liu, H., Guo, T., Wang, T., Tang, H., & Sebe, N. (2025). *GraphMLP: A Graph MLP-Like Architecture for 3D Human Pose Estimation*. Pattern Recognition, 158, 110925

    2025

  38. [46]

    Liu, K., Ding, R., Zou, Z., Wang, L., & Tang, W. (2020). A Comprehensive Study of Weight Sharing in Graph Networks for 3D Human Pose Estimation . In Computer Vision – ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part X 16 (pp. 318–334)

    2020

  39. [47]

    Zeng, A., Sun, X., Yang, L., Zhao, N., Liu, M., & Xu, Q. (2021). Learning Skeletal Graph Neural Networks for Hard 3D Pose Estimation . In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 11436–11445)

    2021

  40. [48]

    Zhai, K., Nie, Q., Ouyang, B., Li, X., & Yang, S. (2023). *HopFIR: Hop-wise GraphFormer with Intragroup Joint Refinement for 3D Human Pose Estimation*. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 14985–14995)

    2023

  41. [49]

    Zhu, Y ., Xu, X., Shen, F., Ji, Y ., Gao, L., & Shen, H. T. (2021). *PoseGTAC: Graph Transformer Encoder-Decoder with Atrous Convolution for 3D Human Pose Estimation*. In Proceedings of the IJCAI (pp. 1359–1365). https://doi.org/10.24963/ijcai.2021/188

    2021 doi

  42. [50]

    Zhang, S., Wang, C., Nie, L., Yao, H., Huang, Q., & Tian, Q. (2023). *Learning Enriched Hop-Aware Correlation for Robust 3D Human Pose Estimation*. International Journal of Computer Vision, 131(6), 1566–1583

    2023

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.