Pith. sign in

REVIEW 3 major objections 2 minor 84 references

Foundation Model for Skeleton-Based Human Action Understanding

T0 review · 3 major / 2 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read USDRL, a unified skeleton pretraining framework, achieves state-of-the-art results on 25 benchmarks across 9 action-understanding tasks.

desk verdict The submitted full text is a BCI cybersecurity paper, not the claimed skeleton foundation model, so there is no USDRL content to evaluate. read the letter →

arxiv 2508.12586 v1 pith:B7GW6MC7 submitted 2025-08-18 cs.CV

Hongsong Wang , Wanjiang Weng , Junbo Wang , Fang Zhao , Guo-Sen Xie , Xin Geng , Liang Wang This is my paper
classification cs.CV
keywords skeleton-basedhumanactionunderstandingfoundationmodeldenserepresentationlearningself-supervisedmulti-viewconsistencymulti-modalpretrainingTransformerrecognition
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

This paper's abstract introduces USDRL (Unified Skeleton-based Dense Representation Learning), a framework intended to be the first foundation model for skeleton-based human action understanding. It claims that a Transformer-based Dense Spatio-Temporal Encoder, a Multi-Grained Feature Decorrelation module, and Multi-Perspective Consistency Training together learn a dense, task-agnostic representation that outperforms current state-of-the-art methods on 25 benchmarks spanning coarse prediction, dense prediction, and transferred prediction. If true, this would mean one pretrained skeleton model can replace task-specific architectures across most action-understanding workloads. However, the supplied full text of this submission is a different manuscript—an analysis of cybersecurity risks to brain-computer interfaces—so none of the USDRL architecture, training procedure, or experimental results can be examined or verified from the provided material.

What carries the argument

The named machinery is the USDRL framework itself. Its load-bearing parts are the Dense Spatio-Temporal Encoder (DSTE), a dual-stream Transformer that separately encodes temporal dynamics and spatial structure; Multi-Grained Feature Decorrelation (MG-FD), which decorrelates features across temporal, spatial, and instance granularities to cut redundancy; and Multi-Perspective Consistency Training (MPCT), which applies multi-view and multi-modal consistency objectives during self-supervised pretraining. Together, they are meant to produce a reusable dense skeleton representation rather than a task-specific one.

What would settle it

Reproduce USDRL's pretraining and fine-tuning on a public skeleton-action benchmark under the same splits and metrics; if the reported state-of-the-art numbers do not appear, the central claim is refuted—and the missing full text already means no such reproduction is possible from this submission.

Watch

Extended reading notes

Core claim

Dense, task-agnostic representation learning for skeletons is the stated goal. USDRL comprises a Dense Spatio-Temporal Encoder (DSTE) with parallel streams for temporal dynamics and spatial structure; Multi-Grained Feature Decorrelation (MG-FD) that reduces redundancy across temporal, spatial, and instance domains; and Multi-Perspective Consistency Training (MPCT), which uses multi-view and multi-modal self-supervision to favor high-level semantics over low-level discrepancies. The paper claims this combination significantly outperforms state-of-the-art methods on 25 benchmarks over 9 tasks, including dense prediction tasks. Because the submitted full text is a different paper, none of this

Load-bearing premise

The claim rests on the assumption that the 25-benchmark, 9-task comparison is fair (matched pretraining and fine-tuning, no dataset-specific tuning) and that the self-supervised training truly produces a task-agnostic dense representation; neither is checkable because the supplied full text is a different manuscript.

Editorial extensions

If this is right

  • If USDRL works as claimed, a single pretrained skeleton encoder could be fine-tuned for coarse tasks (action recognition), dense tasks (segmentation, detection), and transferred tasks with minimal task-specific surgery.
  • Skeleton-based action understanding would join the pretrain-then-finetune paradigm that already dominates image and language modeling, letting researchers share a common backbone instead of training per-task models from scratch.
  • The explicit emphasis on dense prediction tasks could push the field beyond clip-level classification toward frame-level and joint-level understanding, where the abstract claims the largest gaps are.
  • State-of-the-art results across 25 benchmarks would imply that multi-view and multi-modal self-supervision, plus feature decorrelation, are the right inductive biases for skeleton data.

Reading between the lines

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

  • The submission as it stands is unverifiable: the supplied full text is a different manuscript on BCI cybersecurity, so the reader cannot check the architecture, training details, or benchmark tables for USDRL.
  • If the full USDRL paper surfaces, the most informative test would be whether removing MG-FD or MPCT degrades dense and transferred tasks more than coarse ones; the abstract predicts exactly that, because those modules are presented as the source of high-level, task-agnostic features.
  • The paper's framing invites a comparison with foundation models in NLP and vision: if skeleton encoders can be pretrained once and reused, the bottleneck for human action understanding shifts from architecture design to data collection and evaluation protocols, especially for dense tasks.
  • A truly task-agnostic dense representation would also benefit downstream embodied AI systems, such as humanoid robot control and human-robot interaction, which the abstract itself lists as motivation but does not develop.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 2 minor

Summary. The submission, arXiv:2508.12586, presents in its abstract a framework called USDRL (Unified Skeleton-based Dense Representation Learning) for skeleton-based human action understanding. The abstract claims a Transformer-based Dense Spatio-Temporal Encoder, Multi-Grained Feature Decorrelation, and Multi-Perspective Consistency Training, and asserts state-of-the-art results across 25 benchmarks and 9 tasks. However, the supplied full text is an unrelated manuscript titled 'Cyber Risks to Next-Gen Brain-Computer Interfaces: Analysis and Recommendations' that contains no description of USDRL, no methods, no experiments, no benchmark tables, and no results. The central claims of the paper are therefore entirely unsupported by the submitted material.

Significance. If the claimed USDRL framework existed as described and achieved state-of-the-art performance across coarse, dense, and transferred skeleton-based action understanding tasks on 25 benchmarks, it would be a substantial contribution to the field. A single pretrained skeleton foundation model with dense representations and consistency training could indeed broaden the scope of skeleton-based action understanding. However, none of these contributions is present in the full text submitted for review. There are no architectural details, training objectives, experimental protocols, comparison tables, or error analyses. Consequently, the significance of the work cannot be assessed; the submission provides no verifiable evidence for any of its central claims.

major comments (3)
  1. [Full Text (all sections)] The submitted full text (pp. 1–24) is a completely different manuscript on cybersecurity risks to brain-computer interfaces. It contains no mention of USDRL, DSTE, MG-FD, MPCT, skeleton-based action understanding, or any of the 25 benchmarks listed in the abstract. The abstract's technical claims are therefore unsupported by any accompanying methods, derivations, or experimental results. This is a load-bearing omission: there is no way to evaluate the correctness, novelty, or empirical validity of the proposed approach.
  2. [Abstract, empirical claim] The abstract states that USDRL 'significantly outperforms the current state-of-the-art methods' on 25 benchmarks across 9 tasks, but no results, baselines, protocol descriptions, or statistical significance measures appear anywhere in the supplied material. Without tables, error bars, or comparison details, this empirical claim cannot be checked or validated.
  3. [Abstract, MPCT module] The claim that Multi-Perspective Consistency Training (MPCT) with multi-view and multi-modal self-supervised consistency produces a task-agnostic dense representation is a core novelty assertion. The full text provides no derivation, architectural specification, training objective, or ablation study for this component. The assertion is therefore unsupported and not amenable to technical review.
minor comments (2)
  1. [Abstract] Typographical error: 'across across 9 skeleton-based action understanding tasks' should read 'across 9...'.
  2. [Full Text] The full text contains formatting artifacts such as 'Error! Reference source not found.' and duplicate table numbering (two tables labeled Table 3). While not relevant to the abstract's claims, these issues are presentation defects.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity identifiable: the supplied full text is a different manuscript, so the USDRL claims are unverifiable rather than circular.

full rationale

The submitted full text is not the USDRL manuscript described in the abstract. It is a BCI cybersecurity paper by Schroder et al. The abstract claims a skeleton-based foundation model with DSTE, MG-FD, and MPCT modules and state-of-the-art results on 25 benchmarks across 9 tasks, but none of these components, experiments, result tables, or derivations appear in the supplied body. I therefore cannot exhibit any equation or fitted parameter that reduces a prediction to its input; there is no derivation chain to walk. The absence of the actual method and experiments is a serious missing-support / omitted-evidence problem, and it is flagged here per the reviewing rule, but it is not circularity in the sense of a claimed derivation being equivalent to its own inputs. Under the hard rule that circularity must be exhibited with specific quotations and reductions, the honest finding is that no circular step can be identified from the available material.

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

Because the supplied full text is an unrelated BCI manuscript, no parameter values, derivations, or entities can be audited. The axioms listed are the assumptions visible in the abstract.

assumptions (3)
  • domain assumption Skeletons are a modality- and device-agnostic representation sufficient for diverse action understanding tasks.
    Stated in the abstract as motivation; no supporting evidence is present in the supplied text.
  • domain assumption Multi-view and multi-modal self-supervised consistency training improves task-agnostic representation quality.
    Claimed through the MPCT module in the abstract; not demonstrated in the supplied text.
  • domain assumption The reported benchmark comparisons are fair and the state-of-the-art numbers are accurate.
    No protocol details, dataset splits, or error bars are present in the supplied text; the central state-of-the-art claim depends on this assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Foundation Model for Skeleton-Based Human Action Understanding." pith.science (2026). https://pith.science/paper/B7GW6MC7

@misc{pith2026250812586,
  author       = {Pith},
  title        = {Pith review of: Foundation Model for Skeleton-Based Human Action Understanding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/B7GW6MC7}},
  note         = {Machine review of arXiv:2508.12586}
}
read the original abstract

Human action understanding serves as a foundational pillar in the field of intelligent motion perception. Skeletons serve as a modality- and device-agnostic representation for human modeling, and skeleton-based action understanding has potential applications in humanoid robot control and interaction. \RED{However, existing works often lack the scalability and generalization required to handle diverse action understanding tasks. There is no skeleton foundation model that can be adapted to a wide range of action understanding tasks}. This paper presents a Unified Skeleton-based Dense Representation Learning (USDRL) framework, which serves as a foundational model for skeleton-based human action understanding. USDRL consists of a Transformer-based Dense Spatio-Temporal Encoder (DSTE), Multi-Grained Feature Decorrelation (MG-FD), and Multi-Perspective Consistency Training (MPCT). The DSTE module adopts two parallel streams to learn temporal dynamic and spatial structure features. The MG-FD module collaboratively performs feature decorrelation across temporal, spatial, and instance domains to reduce dimensional redundancy and enhance information extraction. The MPCT module employs both multi-view and multi-modal self-supervised consistency training. The former enhances the learning of high-level semantics and mitigates the impact of low-level discrepancies, while the latter effectively facilitates the learning of informative multimodal features. We perform extensive experiments on 25 benchmarks across across 9 skeleton-based action understanding tasks, covering coarse prediction, dense prediction, and transferred prediction. Our approach significantly outperforms the current state-of-the-art methods. We hope that this work would broaden the scope of research in skeleton-based action understanding and encourage more attention to dense prediction tasks.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

84 extracted references · 79 canonical work pages

  1. [1]

    11em plus .33em minus .07em 4000 4000 100 4000 4000 500 `\.=1000 = #1 \@IEEEnotcompsoconly \@IEEEcompsoconly #1 * [1] 0pt [0pt][0pt] #1 * [1] 0pt [0pt][0pt] #1 * \| ** #1 \@IEEEauthorblockNstyle \@IEEEcompsocnotconfonly \@IEEEauthorblockAstyle \@IEEEcompsocnotconfonly \@IEEEcompsocconfonly \@IEEEauthordefaulttextstyle \@IEEEcompsocnotconfonly \@IEEEauthor...

  2. [2]

    Y. Du, W. Wang, and L. Wang, ``Hierarchical recurrent neural network for skeleton based action recognition,'' in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2015, pp. 1110--1118

    2015

  3. [3]

    Y. Du, Y. Fu, and L. Wang, ``Representation learning of temporal dynamics for skeleton-based action recognition,'' IEEE Transactions on Image Processing, vol. 25, no. 7, pp. 3010--3022, 2016

    2016

  4. [4]

    J. Liu, A. Shahroudy, D. Xu, and G. Wang, ``Spatio-temporal lstm with trust gates for 3d human action recognition,'' in European Conference on Computer Vision. 1em plus 0.5em minus 0.4em Springer, 2016, pp. 816--833

    2016

  5. [5]

    Wang and L

    H. Wang and L. Wang, ``Modeling temporal dynamics and spatial configurations of actions using two-stream recurrent neural networks,'' in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 499--508

    work page 2017

  6. [6]

    ------, ``Beyond joints: Learning representations from primitive geometries for skeleton-based action recognition and detection,'' IEEE Transactions on Image Processing, vol. 27, no. 9, pp. 4382--4394, 2018

    work page 2018

  7. [7]

    S. Yan, Y. Xiong, and D. Lin, ``Spatial temporal graph convolutional networks for skeleton-based action recognition,'' in Proceedings of the AAAI conference on artificial intelligence, vol. 32, no. 1, 2018

    work page 2018

  8. [8]

    L. Shi, Y. Zhang, J. Cheng, and H. Lu, ``Two-stream adaptive graph convolutional networks for skeleton-based action recognition,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 12\,026--12\,035

    work page 2019

Show all 84 references
  1. [9]

    Cheng, Y

    K. Cheng, Y. Zhang, X. He, W. Chen, J. Cheng, and H. Lu, ``Skeleton-based action recognition with shift graph convolutional network,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 183--192

    2020

  2. [10]

    Z. Liu, H. Zhang, Z. Chen, Z. Wang, and W. Ouyang, ``Disentangling and unifying graph convolutions for skeleton-based action recognition,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 143--152

    2020

  3. [11]

    Y.-F. Song, Z. Zhang, C. Shan, and L. Wang, ``Constructing stronger and faster baselines for skeleton-based action recognition,'' IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 2, pp. 1474--1488, 2022

    2022

  4. [12]

    H. Yan, Y. Liu, Y. Wei, Z. Li, G. Li, and L. Lin, ``Skeletonmae: graph-based masked autoencoder for skeleton sequence pre-training,'' in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 5606--5618

    2023

  5. [13]

    Y. Mao, J. Deng, W. Zhou, Y. Fang, W. Ouyang, and H. Li, ``Masked motion predictors are strong 3d action representation learners,'' in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 10\,181--10\,191

    2023

  6. [14]

    S. Yang, J. Liu, S. Lu, E. M. Hwa, Y. Hu, and A. C. Kot, ``Self-supervised 3d action representation learning with skeleton cloud colorization,'' IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023

    2023

  7. [15]

    Zhang, Y

    H. Zhang, Y. Hou, W. Zhang, and W. Li, ``Contrastive positive mining for unsupervised 3d action representation learning,'' in European Conference on Computer Vision. 1em plus 0.5em minus 0.4em Springer, 2022, pp. 36--51

    2022

  8. [16]

    T. Guo, H. Liu, Z. Chen, M. Liu, T. Wang, and R. Ding, ``Contrastive learning from extremely augmented skeleton sequences for self-supervised action recognition,'' in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 36, no. 1, 2022, pp. 762--770

    2022

  9. [17]

    J. Liu, A. Shahroudy, G. Wang, L.-Y. Duan, and A. C. Kot, `` SSNet: S cale selection network for online 3d action prediction,'' in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 8349--8358

    2018

  10. [18]

    Awais, M

    M. Awais, M. Naseer, S. Khan, R. M. Anwer, H. Cholakkal, M. Shah, M.-H. Yang, and F. S. Khan, ``Foundation models defining a new era in vision: a survey and outlook,'' IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025

    2025

  11. [19]

    W. Weng, H. Wang, J. Wang, L. He, and G.-S. Xie, `` USDRL: U nified skeleton-based dense representation learning with multi-grained feature decorrelation,'' in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 39, no. 8, 2025, pp. 8332--8340

    2025

  12. [20]

    W. Zhu, X. Ma, Z. Liu, L. Liu, W. Wu, and Y. Wang, ``Motionbert: A unified perspective on learning human motion representations,'' in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 15\,085--15\,099

    2023

  13. [21]

    L. Li, M. Wang, B. Ni, H. Wang, J. Yang, and W. Zhang, ``3d human action representation learning via cross-view consistency pursuit,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 4741--4750

    2021

  14. [22]

    T. Guo, M. Liu, H. Liu, G. Wang, and W. Li, ``Improving self-supervised action recognition from extremely augmented skeleton sequences,'' Pattern Recognition, vol. 150, p. 110333, 2024

    2024

  15. [23]

    Y. Mao, W. Zhou, Z. Lu, J. Deng, and H. Li, ``Cmd: Self-supervised 3d action representation learning with cross-modal mutual distillation,'' in European Conference on Computer Vision. 1em plus 0.5em minus 0.4em Springer, 2022, pp. 734--752

    2022

  16. [24]

    Zhang, L

    J. Zhang, L. Lin, and J. Liu, ``Hierarchical consistent contrastive learning for skeleton-based action recognition with growing augmentations,'' in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 3, 2023, pp. 3427--3435

    2023

  17. [25]

    Zbontar, L

    J. Zbontar, L. Jing, I. Misra, Y. LeCun, and S. Deny, ``Barlow twins: Self-supervised learning via redundancy reduction,'' in International Conference on Machine Learning. 1em plus 0.5em minus 0.4em PMLR, 2021, pp. 12\,310--12\,320

    2021

  18. [26]

    Bardes, J

    A. Bardes, J. Ponce, and Y. LeCun, ``Vicreg: Variance-invariance-covariance regularization for self-supervised learning,'' in International Conference on Learning Representations, 2022

    2022

  19. [27]

    Y. Chen, L. Zhao, J. Yuan, Y. Tian, Z. Xia, S. Geng, L. Han, and D. N. Metaxas, ``Hierarchically self-supervised transformer for human skeleton representation learning,'' in European Conference on Computer Vision. 1em plus 0.5em minus 0.4em Springer, 2022, pp. 185--202

    2022

  20. [28]

    Zhang, L

    J. Zhang, L. Lin, and J. Liu, ``Prompted contrast with masked motion modeling: Towards versatile 3d action representation learning,'' in Proceedings of the ACM International Conference on Multimedia, 2023, pp. 7175--7183

    2023

  21. [29]

    J. Sun, L. Huang, H. Wang, C. Zheng, J. Qiu, M. T. Islam, E. Xie, B. Zhou, L. Xing, A. Chandrasekaran et al., ``Localization and recognition of human action in 3d using transformers,'' Communications Engineering, vol. 3, no. 1, p. 125, 2024

    2024

  22. [30]

    L. Xu, Q. Wang, X. Lin, and L. Yuan, ``An efficient framework for few-shot skeleton-based temporal action segmentation,'' Computer Vision and Image Understanding, vol. 232, p. 103707, 2023

    2023

  23. [31]

    Li, K.-Y

    Y.-H. Li, K.-Y. Liu, S.-L. Liu, L. Feng, and H. Qiao, ``Involving distinguished temporal graph convolutional networks for skeleton-based temporal action segmentation,'' IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 1, pp. 647--660, 2023

    2023

  24. [32]

    D. Yang, Y. Wang, A. Dantcheva, Q. Kong, L. Garattoni, G. Francesca, and F. Bremond, ``Lac-latent action composition for skeleton-based action segmentation,'' in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 13\,679--13\,690

    2023

  25. [33]

    S. W. Hyder, M. Usama, A. Zafar, M. Naufil, F. J. Fateh, A. Konin, M. Z. Zia, and Q.-H. Tran, ``Action segmentation using 2d skeleton heatmaps and multi-modality fusion,'' in IEEE International Conference on Robotics and Automation. 1em plus 0.5em minus 0.4em IEEE, 2024, pp. 1...

    2024

  26. [34]

    H. Ji, B. Chen, X. Xu, W. Ren, Z. Wang, and H. Liu, ``Language-assisted skeleton action understanding for skeleton-based temporal action segmentation,'' in European Conference on Computer Vision. 1em plus 0.5em minus 0.4em Springer, 2024, pp. 400--417

    2024

  27. [35]

    M. S. Aliakbarian, F. S. Saleh, M. Salzmann, B. Fernando, L. Petersson, and L. Andersson, ``Encouraging lstms to anticipate actions very early,'' in IEEE International Conference on Computer Vision, vol. 1, no. 2, 2017

    2017

  28. [36]

    Y. Kong, S. Gao, B. Sun, and Y. Fu, ``Action prediction from videos via memorizing hard-to-predict samples,'' in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 32, no. 1, 2018

    2018

  29. [37]

    Wang, J.-F

    X. Wang, J.-F. Hu, J.-H. Lai, J. Zhang, and W.-S. Zheng, ``Progressive teacher-student learning for early action prediction,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 3556--3565

    2019

  30. [38]

    X. Wang, Z. Fang, X. Li, X. Li, C. Chen, and M. Liu, ``Skeleton-in-context: Unified skeleton sequence modeling with in-context learning,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 2436--2446

    2024

  31. [39]

    S. Sun, D. Liu, J. Dong, X. Qu, J. Gao, X. Yang, X. Wang, and M. Wang, ``Unified multi-modal unsupervised representation learning for skeleton-based action understanding,'' in Proceedings of the ACM International Conference on Multimedia, 2023, pp. 2973--2984

    2023

  32. [40]

    L. G. Foo, T. Li, H. Rahmani, Q. Ke, and J. Liu, ``Unified pose sequence modeling,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 13\,019--13\,030

    2023

  33. [41]

    H. Qu, Y. Cai, and J. Liu, `` LLMs are good action recognizers,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 18\,395--18\,406

    2024

  34. [42]

    H. Wang, X. Ma, J. Kuang, and J. Gui, ``Heterogeneous skeleton-based action representation learning,'' in Proceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 19\,154--19\,164

    2025

  35. [43]

    Ermolov, A

    A. Ermolov, A. Siarohin, E. Sangineto, and N. Sebe, ``Whitening for self-supervised representation learning,'' in International Conference on Machine Learning. 1em plus 0.5em minus 0.4em PMLR, 2021, pp. 3015--3024

    2021

  36. [44]

    H. Xu, G. Ghosh, P.-Y. Huang, D. Okhonko, A. Aghajanyan, F. Metze, L. Zettlemoyer, and C. Feichtenhofer, ``Videoclip: Contrastive pre-training for zero-shot video-text understanding,'' in Proceedings of Conference on Empirical Methods in Natural Language Processing, 2021, pp. ...

    2021

  37. [45]

    D. Li, J. Li, H. Li, J. C. Niebles, and S. C. Hoi, ``Align and prompt: Video-and-language pre-training with entity prompts,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 4953--4963

    2022

  38. [46]

    H. Xue, T. Hang, Y. Zeng, Y. Sun, B. Liu, H. Yang, J. Fu, and B. Guo, ``Advancing high-resolution video-language representation with large-scale video transcriptions,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 5036--5045

    2022

  39. [47]

    L. Wang, B. Huang, Z. Zhao, Z. Tong, Y. He, Y. Wang, Y. Wang, and Y. Qiao, ``Videomae v2: Scaling video masked autoencoders with dual masking,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 14\,549--14\,560

    2023

  40. [48]

    R. Wang, D. Chen, Z. Wu, Y. Chen, X. Dai, M. Liu, Y.-G. Jiang, L. Zhou, and L. Yuan, ``Bevt: Bert pretraining of video transformers,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 14\,733--14\,743

    2022

  41. [49]

    Xu and Y

    Y. Xu and Y. Fu, `` Sports-Traj: A unified trajectory generation model for multi-agent movement in sports,'' in The Thirteenth International Conference on Learning Representations, 2025

    2025

  42. [50]

    Zhang, D

    M. Zhang, D. Jin, C. Gu, F. Hong, Z. Cai, J. Huang, C. Zhang, X. Guo, L. Yang, Y. He et al., ``Large motion model for unified multi-modal motion generation,'' in European Conference on Computer Vision. 1em plus 0.5em minus 0.4em Springer, 2024, pp. 397--421

    2024

  43. [51]

    Y. Li, Z. Li, S. Gao, Q. Wang, Q. Hou, and M.-M. Cheng, ``A decoupled spatio-temporal framework for skeleton-based action segmentation,'' arXiv preprint arXiv:2312.05830, 2023

    2023 arXiv

  44. [52]

    Shahroudy, J

    A. Shahroudy, J. Liu, T.-T. Ng, and G. Wang, ``Ntu rgb+ d: A large scale dataset for 3d human activity analysis,'' in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 1010--1019

    2016

  45. [53]

    J. Liu, A. Shahroudy, M. Perez, G. Wang, L.-Y. Duan, and A. C. Kot, ``Ntu rgb+ d 120: A large-scale benchmark for 3d human activity understanding,'' IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 42, no. 10, pp. 2684--2701, 2019

    2019

  46. [54]

    J. Liu, S. Song, C. Liu, Y. Li, and Y. Hu, ``A benchmark dataset and comparison study for multi-modal human action analytics,'' ACM Transactions on Multimedia Computing, Communications, and Applications, vol. 16, no. 2, pp. 1--24, 2020

    2020

  47. [55]

    T. Li, J. Liu, W. Zhang, Y. Ni, W. Wang, and Z. Li, `` UAV-H uman: A large benchmark for human behavior understanding with unmanned aerial vehicles,'' in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 16\,266--16\,275

    2021

  48. [56]

    L. Lin, S. Song, W. Yang, and J. Liu, ``Ms2l: Multi-task self-supervised learning for skeleton based action recognition,'' in Proceedings of the ACM International Conference on Multimedia, 2020, pp. 2490--2498

    2020

  49. [57]

    S. Yang, J. Liu, S. Lu, M. H. Er, and A. C. Kot, ``Skeleton cloud colorization for unsupervised 3d action representation learning,'' in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 13\,423--13\,433

    2021

  50. [58]

    B. Kim, H. J. Chang, J. Kim, and J. Y. Choi, ``Global-local motion transformer for unsupervised skeleton-based action learning,'' in European Conference on Computer Vision. 1em plus 0.5em minus 0.4em Springer, 2022, pp. 209--225

    2022

  51. [59]

    W. Wu, Y. Hua, C. Zheng, S. Wu, C. Chen, and A. Lu, ``Skeletonmae: Spatial-temporal masked autoencoders for self-supervised skeleton action recognition,'' in IEEE International Conference on Multimedia and Expo Workshops. 1em plus 0.5em minus 0.4em IEEE, 2023, pp. 224--229

    2023

  52. [60]

    L. Wu, L. Lin, J. Zhang, Y. Ma, and J. Liu, `` MacDiff: U nified skeleton modeling with masked conditional diffusion,'' in European Conference on Computer Vision. 1em plus 0.5em minus 0.4em Springer, 2024, pp. 110--128

    2024

  53. [61]

    X. Zhu, X. Shu, and J. Tang, ``Motion-aware mask feature reconstruction for skeleton-based action recognition,'' IEEE Transactions on Circuits and Systems for Video Technology, 2024

    2024

  54. [62]

    Y. Zhou, H. Duan, A. Rao, B. Su, and J. Wang, ``Self-supervised action representation learning from partial spatio-temporal skeleton sequences,'' in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 3, 2023, pp. 3825--3833

    2023

  55. [63]

    A. Shah, A. Roy, K. Shah, S. Mishra, D. Jacobs, A. Cherian, and R. Chellappa, ``Halp: Hallucinating latent positives for skeleton-based self-supervised learning of actions,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, June 2023, pp. 1...

    2023

  56. [64]

    J. Dong, S. Sun, Z. Liu, S. Chen, B. Liu, and X. Wang, ``Hierarchical contrast for unsupervised skeleton-based action representation learning,'' in Proceedings of the AAAI Conference on Artificial Intelligence, 2023

    2023

  57. [65]

    S. Guan, X. Yu, W. Huang, G. Fang, and H. Lu, ``Dmmg: dual min-max games for self-supervised skeleton-based action recognition,'' IEEE Transactions on Image Processing, vol. 33, pp. 395--407, 2023

    2023

  58. [66]

    J. Hu, Y. Hou, Z. Guo, and J. Gao, ``Global and local contrastive learning for self-supervised skeleton-based action recognition,'' IEEE Transactions on Circuits and Systems for Video Technology, 2024

    2024

  59. [67]

    Wang and Y

    X. Wang and Y. Mu, ``Localized linear temporal dynamics for self-supervised skeleton action recognition,'' IEEE Transactions on Multimedia, 2024

    2024

  60. [68]

    Wu, X.-J

    C. Wu, X.-J. Wu, J. Kittler, T. Xu, S. Ahmed, M. Awais, and Z. Feng, ``Scd-net: Spatiotemporal clues disentanglement network for self-supervised skeleton-based action recognition,'' in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 38, no. 6, 2024, pp. 5949--5957

    2024

  61. [69]

    D. Yang, Y. Wang, A. Dantcheva, L. Garattoni, G. Francesca, and F. Br \'e mond, ``View-invariant skeleton action representation learning via motion retargeting,'' International Journal of Computer Vision, pp. 1--16, 2024

    2024

  62. [70]

    L. Lin, L. Wu, J. Zhang, and J. Liu, ``Idempotent unsupervised representation learning for skeleton-based action recognition,'' in European Conference on Computer Vision. 1em plus 0.5em minus 0.4em Springer, 2024, pp. 75--92

    2024

  63. [71]

    Franco, P

    L. Franco, P. Mandica, B. Munjal, and F. Galasso, ``Hyperbolic self-paced learning for self-supervised skeleton-based action representations,'' in International Conference on Learning Representations, 2023

    2023

  64. [72]

    Y. Hua, W. Wu, C. Zheng, A. Lu, M. Liu, C. Chen, and S. Wu, ``Part aware contrastive learning for self-supervised action recognition,'' in Proceedings of the International Joint Conference on Artificial Intelligence, 2023, pp. 855--863

    2023

  65. [73]

    Z. Jin, Y. Wang, Q. Wang, Y. Shen, and H. Meng, ``Ssrl: Self-supervised spatial-temporal representation learning for 3d action recognition,'' IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 1, pp. 274--285, 2023

    2023

  66. [74]

    L. Lin, J. Zhang, and J. Liu, ``Actionlet-dependent contrastive learning for unsupervised skeleton-based action recognition,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 2363--2372

    2023

  67. [75]

    Y. Zhu, H. Han, Z. Yu, and G. Liu, ``Modeling the relative visual tempo for self-supervised skeleton-based action recognition,'' in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 13\,913--13\,922

    2023

  68. [76]

    M. Wang, X. Li, S. Chen, X. Zhang, L. Ma, and Y. Zhang, ``Learning representations by contrastive spatio-temporal clustering for skeleton-based action recognition,'' IEEE Transactions on Multimedia, vol. 26, pp. 3207--3220, 2023

    2023

  69. [77]

    Plizzari, M

    C. Plizzari, M. Cannici, and M. Matteucci, ``Skeleton-based action recognition via spatial and temporal transformer networks,'' Computer Vision and Image Understanding, vol. 208, p. 103219, 2021

    2021

  70. [78]

    F. M. Thoker, H. Doughty, and C. G. Snoek, ``Skeleton-contrastive 3d action representation learning,'' in Proceedings of the ACM International Conference on Multimedia, 2021, pp. 1655--1663

    2021

  71. [79]

    T. Li, J. Liu, W. Zhang, and L. Duan, ``Hard- Net: H ardness-aware discrimination network for 3d early activity prediction,'' in European Conference on Computer Vision. 1em plus 0.5em minus 0.4em Springer, 2020, pp. 420--436

    2020

  72. [80]

    P. Wang, J. Wen, C. Si, Y. Qian, and L. Wang, ``Contrast-reconstruction representation learning for self-supervised skeleton-based action recognition,'' IEEE Transactions on Image Processing, vol. 31, pp. 6224--6238, 2022

    2022

  73. [81]

    Y. Kong, Z. Tao, and Y. Fu, ``Deep sequential context networks for action prediction,'' in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 1473--1481

    2017

  74. [82]

    Hu, W.-S

    J.-F. Hu, W.-S. Zheng, L. Ma, G. Wang, J. Lai, and J. Zhang, ``Early action prediction by soft regression,'' IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 41, no. 11, pp. 2568--2583, 2018

    2018

  75. [83]

    Y. A. Farha and J. Gall, `` MS-TCN: M ulti-stage temporal convolutional network for action segmentation,'' in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 3575--3584

    2019

  76. [84]

    Y. Li, Z. Dong, K. Liu, L. Feng, L. Hu, J. Zhu, L. Xu, S. Liu et al., ``Efficient two-step networks for temporal action segmentation,'' Neurocomputing, vol. 454, pp. 373--381, 2021

    2021

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.