Pith. sign in

REVIEW 3 major objections 2 minor 43 references

Exploring Adaptive Masked Reconstruction for Self-Supervised Skeleton-Based Action Recognition

T0 review · 3 major / 2 minor · reviewed 2026-06-27 · grok-4.3

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

Pith's one-line read Adaptive Masked Reconstruction speeds up self-supervised pre-training for skeleton action recognition while improving accuracy over prior methods.

desk verdict AMR decouples the decoder for larger patches and adds an adaptive guidance module to focus on high-motion skeleton regions, delivering faster pretraining and accuracy gains on standard benchmarks. read the letter →

arxiv 2606.11450 v1 pith:67GXILNH submitted 2026-06-09 cs.CV

Shengkai Sun , Zhiyong Cheng , Zefan Zhang , Jianfeng Dong , Zhihui Li , Meng Wang This is my paper · ORCID
classification cs.CV
keywords self-supervisedlearningskeletonactionrecognitionmaskedreconstructionadaptiveguidancepre-trainingaccelerationmotioninformativeness
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 introduces Adaptive Masked Reconstruction (AMR) to overcome two issues in existing masked skeleton models: long training times from predicting many small patches and distraction from uniform treatment of all regions. AMR decouples the decoder to allow larger patches, cutting reconstruction complexity. An adaptive guidance module then highlights high-motion areas within those patches to keep the task focused on key action patterns. Tests on standard datasets show both faster training and higher recognition rates than current approaches.

What carries the argument

The adaptive guidance module, which identifies regions of high motion informativeness to direct the model's attention within larger reconstructed patches.

What would settle it

Running the experiments and finding that AMR does not reduce pre-training time or fails to exceed prior accuracy on NTU RGB+D 60 would disprove the central claim.

Watch

Extended reading notes

Core claim

AMR decouples the decoder from the encoder to enable prediction of larger spatiotemporal patches, which reduces reconstruction complexity, and adds an adaptive guidance module that identifies high motion informativeness regions to guide focus and alleviate difficulty, leading to substantially accelerated pre-training and improved downstream recognition accuracy that surpasses state-of-the-art methods on NTU RGB+D 60, NTU RGB+D 120, and PKU-MMD datasets.

Load-bearing premise

That the adaptive guidance module can reliably identify regions of high motion informativeness and that focusing reconstruction on those regions will reduce difficulty without discarding semantically important low-motion cues.

Editorial extensions

If this is right

  • Pre-training becomes substantially faster due to larger patch prediction.
  • Downstream action recognition accuracy increases on multiple benchmark datasets.
  • The method outperforms existing state-of-the-art self-supervised approaches.
  • Larger patches become feasible without degrading performance through guided focus.

Reading between the lines

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

  • Similar decoupling and adaptive focus strategies could apply to other self-supervised vision tasks involving complex spatiotemporal data.
  • Emphasizing motion cues might help in distinguishing actions where static poses are ambiguous.
  • Reduced training time could enable more iterative experimentation in developing new skeleton models.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 2 minor

Summary. The paper proposes Adaptive Masked Reconstruction (AMR) for self-supervised skeleton-based action recognition. It decouples the decoder to permit larger spatiotemporal patches and thereby reduce pre-training complexity. An adaptive guidance module is introduced to identify high-motion-informative regions within patches and focus reconstruction on the most discriminative parts. Experiments on NTU RGB+D 60, NTU RGB+D 120, and PKU-MMD report substantially faster pre-training together with higher downstream recognition accuracy that surpasses current state-of-the-art methods.

Significance. If the reported gains are substantiated, the work would be significant for the field: it directly targets the computational bottleneck of predicting large numbers of patches in masked skeleton reconstruction while attempting to steer learning toward motion-critical regions. Multi-dataset evaluation is a strength; reproducible code or detailed ablations would further strengthen the contribution.

major comments (3)
  1. [Section 3.2] Section 3.2 (adaptive guidance module): the central claim that the module 'identifies regions of high motion informativeness' and thereby 'alleviates reconstruction difficulty' without discarding semantically important low-motion cues is load-bearing, yet the manuscript provides no ablation (uniform/random selection baseline), no motion-stratified reconstruction error, and no region-selection visualizations to test whether the selection criterion is reliable or biased.
  2. [Section 4] Section 4 (experiments): accuracy improvements over SOTA are asserted on NTU RGB+D 60/120 and PKU-MMD, but the tables do not report standard deviations across runs, number of random seeds, or statistical significance tests; without these, it is impossible to determine whether the gains are robust or could be explained by variance.
  3. [Section 3.1] Section 3.1 (decoder decoupling): the claim of 'dramatically reducing reconstruction complexity' is central to the efficiency contribution, yet no explicit quantification (patch count before/after, FLOPs, or wall-clock time on the same hardware) is supplied to support the 'substantially accelerates' statement.
minor comments (2)
  1. [Section 3.2] Notation for the motion statistic inside the adaptive guidance module is introduced without an equation or pseudocode; a short formal definition would improve clarity.
  2. [Related Work] The abstract and introduction cite 'existing state-of-the-art methods' but do not name the specific masked-reconstruction baselines being compared; adding these citations in the related-work section would help readers situate the contribution.

Simulated Author's Rebuttal

3 responses · 0 unresolved

Thank you for the constructive feedback on our manuscript. We appreciate the referee's identification of areas where additional evidence and quantification would strengthen the claims. We address each major comment below and commit to revisions that directly incorporate the suggested improvements.

read point-by-point responses

  1. Referee: [Section 3.2] Section 3.2 (adaptive guidance module): the central claim that the module 'identifies regions of high motion informativeness' and thereby 'alleviates reconstruction difficulty' without discarding semantically important low-motion cues is load-bearing, yet the manuscript provides no ablation (uniform/random selection baseline), no motion-stratified reconstruction error, and no region-selection visualizations to test whether the selection criterion is reliable or biased.

    Authors: We agree that the current manuscript lacks direct empirical support for the adaptive guidance module's selection behavior. In the revised version we will add: (1) an ablation comparing the proposed motion-based selection against uniform and random baselines on downstream accuracy and pre-training loss; (2) motion-stratified reconstruction error curves that separate high- versus low-motion regions; and (3) qualitative visualizations of selected patches across sample sequences. These additions will allow readers to assess whether the criterion reliably prioritizes discriminative motion while preserving semantically relevant low-motion information. revision: yes

  2. Referee: [Section 4] Section 4 (experiments): accuracy improvements over SOTA are asserted on NTU RGB+D 60/120 and PKU-MMD, but the tables do not report standard deviations across runs, number of random seeds, or statistical significance tests; without these, it is impossible to determine whether the gains are robust or could be explained by variance.

    Authors: We acknowledge that the absence of variability measures and significance testing limits the strength of the reported gains. We will re-run all main experiments using at least three random seeds, report mean accuracy together with standard deviations in the updated tables, and include paired statistical significance tests (e.g., t-tests) against the strongest baselines. These results will be added to Section 4 and the supplementary material. revision: yes

  3. Referee: [Section 3.1] Section 3.1 (decoder decoupling): the claim of 'dramatically reducing reconstruction complexity' is central to the efficiency contribution, yet no explicit quantification (patch count before/after, FLOPs, or wall-clock time on the same hardware) is supplied to support the 'substantially accelerates' statement.

    Authors: The manuscript text asserts a complexity reduction from decoder decoupling and larger patches but does not supply concrete metrics. In the revision we will insert a dedicated efficiency analysis subsection that reports: (i) the exact number of spatiotemporal patches predicted before versus after decoupling, (ii) FLOPs for the reconstruction head on identical hardware, and (iii) measured wall-clock pre-training time for the same number of epochs. A corresponding table will be added to Section 3.1 or 4. revision: yes

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: empirical method proposal without derivational claims

full rationale

The paper proposes AMR as a practical framework (decoder decoupling plus adaptive guidance module) that builds on existing masked reconstruction techniques. No equations, first-principles derivations, or predictions are presented that could reduce to fitted inputs or self-citations by construction. All performance claims rest on empirical results across NTU and PKU-MMD datasets rather than any definitional or fitted equivalence. This is the standard non-circular outcome for an applied CV method paper.

Assumptions & free parameters 0 free parameters · 1 assumptions · 1 invented entities

Paper introduces new module and design choice without independent evidence outside the claimed experiments; relies on standard self-supervised learning assumptions.

assumptions (1)
  • domain assumption Masked reconstruction serves as an effective self-supervised pretext task for learning skeleton action representations.
    Invoked as the foundation for the entire pre-training approach.
invented entities (1)
  • Adaptive guidance module
    purpose: Identifies high-motion regions within patches to guide reconstruction focus.
    New component introduced to address reconstruction difficulty; no independent evidence provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Exploring Adaptive Masked Reconstruction for Self-Supervised Skeleton-Based Action Recognition." pith.science (2026). https://pith.science/paper/67GXILNH

@misc{pith2026260611450,
  author       = {Pith},
  title        = {Pith review of: Exploring Adaptive Masked Reconstruction for Self-Supervised Skeleton-Based Action Recognition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/67GXILNH}},
  note         = {Machine review of arXiv:2606.11450}
}
read the original abstract

Recently, masked skeleton reconstruction models have emerged as strong action representation learners, driving significant progress in self-supervised skeleton-based action recognition. However, existing state-of-the-art methods must predict an exceedingly large number of spatiotemporal patches, significantly prolonging training time. Besides, by treating all spatiotemporal regions equally during reconstruction, these models are distracted from learning the critical motion patterns that underlie action semantics. To address these challenges, we propose Adaptive Masked Reconstruction (AMR), a faster and stronger pre-training framework. We first decouple the decoder from the encoder, enabling flexible prediction of larger spatiotemporal patches and dramatically reducing reconstruction complexity. Given that larger patches contain more complex information, which is challenging to predict and consequently degrades performance, we accordingly introduce an adaptive guidance module. This module identifies regions of high motion informativeness, guiding the model to focus on the most discriminative parts of each patch and alleviating reconstruction difficulty. Experiments on NTU RGB+D 60, NTU RGB+D 120, and PKU-MMD datasets demonstrate that AMR not only accelerates pre-training substantially but also improves downstream recognition accuracy, surpassing current state-of-the-art approaches.

Figures

Figures reproduced from arXiv: 2606.11450 by the authors.

Figure 1
Figure 1. Top-1 accuracy of linear evaluation and training time [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The framework of our proposed AMR. the presence of a missing patch to be predicted. The visi￾ble patch features Hv and the mask tokens M are concate￾nated to form the decoder input He = [Hv∥M] ∈ R N×Ce . Notably, the number of obtained patches of the skeleton se￾quence can be very large. Such a long input sequence leads to a high computational cost for the self-attention module in the decoder. Moreover, unlike the l… view at source ↗
Figure 3
Figure 3. Impact of patch size on model performance. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Visualization of joint importance for the “take off a hat” action. Joints with high weight are depicted in blue. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 3 canonical work pages

  1. [1]

    S-jepa: A joint embedding predictive architecture for skeletal action recog- nition

    Mohamed Abdelfattah and Alexandre Alahi. S-jepa: A joint embedding predictive architecture for skeletal action recog- nition. InEuropean Conference on Computer Vision, pages 367–384. Springer, 2024. 3, 5, 6, 11

    2024

  2. [2]

    Vi- creg: Variance-invariance-covariance regularization for self- supervised learning

    Adrien Bardes, Jean Ponce, and Yann Lecun. Vi- creg: Variance-invariance-covariance regularization for self- supervised learning. InICLR 2022-International Conference on Learning Representations, 2022. 1

    2022

  3. [3]

    Hi- erarchically self-supervised transformer for human skeleton representation learning

    Yuxiao Chen, Long Zhao, Jianbo Yuan, Yu Tian, Zhaoyang Xia, Shijie Geng, Ligong Han, and Dimitris N Metaxas. Hi- erarchically self-supervised transformer for human skeleton representation learning. InEuropean Conference on Com- puter Vision, pages 185–202, 2022. 6

    2022

  4. [4]

    Skeleton-based action recognition with shift graph convolutional network

    Ke Cheng, Yifan Zhang, Xiangyu He, Weihan Chen, Jian Cheng, and Hanqing Lu. Skeleton-based action recognition with shift graph convolutional network. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 183–192, 2020. 1

    2020

  5. [5]

    Hierarchical transformer: Unsupervised representation learning for skeleton-based hu- man action recognition

    Yi-Bin Cheng, Xipeng Chen, Junhong Chen, Pengxu Wei, Dongyu Zhang, and Liang Lin. Hierarchical transformer: Unsupervised representation learning for skeleton-based hu- man action recognition. In2021 IEEE International Con- ference on Multimedia and Expo (ICME), pages 1–6, 2021. 3

    2021

  6. [6]

    Hierarchical contrast for unsu- pervised skeleton-based action representation learning

    Jianfeng Dong, Shengkai Sun, Zhonglin Liu, Shujie Chen, Baolong Liu, and Xun Wang. Hierarchical contrast for unsu- pervised skeleton-based action representation learning. In Proceedings of the AAAI Conference on Artificial Intelli- gence, pages 525–533, 2023. 1, 6, 7

    2023

  7. [7]

    An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl- vain Gelly, et al. An image is worth 16x16 words: Trans- formers for image recognition at scale.arXiv preprint arXiv:2010.11929, 2020. 2

    work page Pith review arXiv 2010

  8. [8]

    Hyperbolic self-paced learning for self-supervised skeleton-based action representations.arXiv preprint arXiv:2303.06242, 2023

    Luca Franco, Paolo Mandica, Bharti Munjal, and Fabio Galasso. Hyperbolic self-paced learning for self-supervised skeleton-based action representations.arXiv preprint arXiv:2303.06242, 2023. 6

    work page arXiv 2023

Show all 43 references
  1. [9]

    Rethinking masked data reconstruction pretraining for strong 3d action representation learning

    Tao Gong, Qi Chu, Bin Liu, and Nenghai Yu. Rethinking masked data reconstruction pretraining for strong 3d action representation learning. InProceedings of the AAAI Confer- ence on Artificial Intelligence, pages 3149–3157, 2025. 3, 6, 7, 11

    2025

  2. [10]

    Contrastive learning from ex- tremely augmented skeleton sequences for self-supervised action recognition

    Tianyu Guo, Hong Liu, Zhan Chen, Mengyuan Liu, Tao Wang, and Runwei Ding. Contrastive learning from ex- tremely augmented skeleton sequences for self-supervised action recognition. InProceedings of the AAAI Conference on Artificial Intelligence, pages 762–770, 2022. 2, 6

    2022

  3. [11]

    Momentum contrast for unsupervised visual rep- resentation learning

    Kaiming He, Haoqi Fan, Yuxin Wu, Saining Xie, and Ross Girshick. Momentum contrast for unsupervised visual rep- resentation learning. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 9729–9738, 2020. 1

    2020

  4. [12]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Doll´ar, and Ross Girshick. Masked autoencoders are scalable vision learners. InProceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition, pages 16000– 16009, 2022. 2

    2022

  5. [13]

    Part aware contrastive learning for self-supervised action recognition

    Yilei Hua, Wenhan Wu, Ce Zheng, Aidong Lu, Mengyuan Liu, Chen Chen, and Shiqian Wu. Part aware contrastive learning for self-supervised action recognition. InInterna- tional Joint Conference on Artificial Intelligence, 2023. 3

    2023

  6. [14]

    Global-local motion transformer for unsupervised skeleton-based action learning

    Boeun Kim, Hyung Jin Chang, Jungho Kim, and Jin Young Choi. Global-local motion transformer for unsupervised skeleton-based action learning. InEuropean Conference on Computer Vision, pages 209–225, 2022. 2, 3

    2022

  7. [15]

    Unsupervised feature learning of human actions as trajectories in pose embedding manifold

    Jogendra Nath Kundu, Maharshi Gor, Phani Krishna Uppala, and Venkatesh Babu Radhakrishnan. Unsupervised feature learning of human actions as trajectories in pose embedding manifold. In2019 IEEE Winter Conference on Applications of Computer Vision (WACV), pages 1459–1467, 2019. 3

    2019

  8. [16]

    3d human action rep- resentation learning via cross-view consistency pursuit

    Linguo Li, Minsi Wang, Bingbing Ni, Hang Wang, Jiancheng Yang, and Wenjun Zhang. 3d human action rep- resentation learning via cross-view consistency pursuit. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 4741–4750, 2021. 3, 7

    2021

  9. [17]

    Ms2l: Multi-task self-supervised learning for skeleton based action recognition

    Lilang Lin, Sijie Song, Wenhan Yang, and Jiaying Liu. Ms2l: Multi-task self-supervised learning for skeleton based action recognition. InProceedings of the 28th ACM International Conference on Multimedia, pages 2490–2498, 2020. 2, 7

    2020

  10. [18]

    Actionlet- dependent contrastive learning for unsupervised skeleton- based action recognition

    Lilang Lin, Jiahang Zhang, and Jiaying Liu. Actionlet- dependent contrastive learning for unsupervised skeleton- based action recognition. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2363–2372, 2023. 3

    2023

  11. [19]

    Ntu rgb+d 120: A large- scale benchmark for 3d human activity understanding.IEEE Transactions on Pattern Analysis and Machine Intelligence, 42(10):2684–2701, 2019

    Jun Liu, Amir Shahroudy, Mauricio Perez, Gang Wang, Ling-Yu Duan, and Alex C Kot. Ntu rgb+d 120: A large- scale benchmark for 3d human activity understanding.IEEE Transactions on Pattern Analysis and Machine Intelligence, 42(10):2684–2701, 2019. 5, 11

    2019

  12. [20]

    A benchmark dataset and comparison study for multi-modal human action analytics.ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM), 16(2):1–24, 2020

    Jiaying Liu, Sijie Song, Chunhui Liu, Yanghao Li, and Yueyu Hu. A benchmark dataset and comparison study for multi-modal human action analytics.ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM), 16(2):1–24, 2020. 5, 11

    2020

  13. [21]

    Cmd: Self-supervised 3d action representation learning with cross-modal mutual distillation

    Yunyao Mao, Wengang Zhou, Zhenbo Lu, Jiajun Deng, and Houqiang Li. Cmd: Self-supervised 3d action representation learning with cross-modal mutual distillation. InEuropean Conference on Computer Vision, pages 734–752. Springer,

  14. [22]

    Masked motion predictors are strong 3d action representation learners

    Yunyao Mao, Jiajun Deng, Wengang Zhou, Yao Fang, Wanli Ouyang, and Houqiang Li. Masked motion predictors are strong 3d action representation learners. InProceedings of the IEEE/CVF International Conference on Computer Vi- sion, pages 10181–10191, 2023. 2, 3, 6, 7, 11

    2023

  15. [23]

    Unsupervised 3d human pose representation with viewpoint and pose disen- tanglement

    Qiang Nie, Ziwei Liu, and Yunhui Liu. Unsupervised 3d human pose representation with viewpoint and pose disen- tanglement. InEuropean Conference on Computer Vision, pages 102–118, 2020. 3

    2020

  16. [24]

    Augmented skeleton based contrastive action learning with momentum lstm for unsupervised action recognition

    Haocong Rao, Shihao Xu, Xiping Hu, Jun Cheng, and Bin Hu. Augmented skeleton based contrastive action learning with momentum lstm for unsupervised action recognition. Information Sciences, 569:90–109, 2021. 3

    2021

  17. [25]

    Halp: Hal- lucinating latent positives for skeleton-based self-supervised learning of actions

    Anshul Shah, Aniket Roy, Ketul Shah, Shlok Mishra, David Jacobs, Anoop Cherian, and Rama Chellappa. Halp: Hal- lucinating latent positives for skeleton-based self-supervised learning of actions. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognit...

    2023

  18. [26]

    Ntu rgb+d: A large scale dataset for 3d human activity anal- ysis

    Amir Shahroudy, Jun Liu, Tian-Tsong Ng, and Gang Wang. Ntu rgb+d: A large scale dataset for 3d human activity anal- ysis. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 1010–1019, 2016. 5, 11

    2016

  19. [27]

    Adversarial self-supervised learn- ing for semi-supervised 3d action recognition

    Chenyang Si, Xuecheng Nie, Wei Wang, Liang Wang, Tie- niu Tan, and Jiashi Feng. Adversarial self-supervised learn- ing for semi-supervised 3d action recognition. InEuropean Conference on Computer Vision, pages 35–51, 2020. 6

    2020

  20. [28]

    Predict & cluster: Unsupervised skeleton based action recognition

    Kun Su, Xiulong Liu, and Eli Shlizerman. Predict & cluster: Unsupervised skeleton based action recognition. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 9631–9640, 2020. 3, 6

    2020

  21. [29]

    Uni- fied multi-modal unsupervised representation learning for skeleton-based action understanding

    Shengkai Sun, Daizong Liu, Jianfeng Dong, Xiaoye Qu, Junyu Gao, Xun Yang, Xun Wang, and Meng Wang. Uni- fied multi-modal unsupervised representation learning for skeleton-based action understanding. InProceedings of the 31st ACM International Conference on Multimedia, pages 29...

    2023

  22. [30]

    Towards efficient general feature prediction in masked skeleton modeling

    Shengkai Sun, Zefan Zhang, Jianfeng Dong, Zhiyong Cheng, Xiaojun Chang, and Meng Wang. Towards efficient general feature prediction in masked skeleton modeling. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 12212–12221, 2025. 2, 3, 6, 7

    2025

  23. [31]

    Skeleton-contrastive 3d action representation learn- ing

    Fida Mohammad Thoker, Hazel Doughty, and Cees GM Snoek. Skeleton-contrastive 3d action representation learn- ing. InProceedings of the 29th ACM International Confer- ence on Multimedia, pages 1655–1663, 2021. 3, 5, 7

    2021

  24. [32]

    Heterogeneous skeleton-based action representation learn- ing

    Hongsong Wang, Xiaoyan Ma, Jidong Kuang, and Jie Gui. Heterogeneous skeleton-based action representation learn- ing. InProceedings of the Computer Vision and Pattern Recognition Conference, pages 19154–19164, 2025. 6, 7

    2025

  25. [33]

    Videomae v2: Scaling video masked autoencoders with dual masking

    Limin Wang, Bingkun Huang, Zhiyu Zhao, Zhan Tong, Yi- nan He, Yi Wang, Yali Wang, and Yu Qiao. Videomae v2: Scaling video masked autoencoders with dual masking. In Proceedings of the IEEE/CVF conference on Computer Vi- sion and Pattern Recognition, pages 14549–14560, 2023. 7

    2023

  26. [34]

    Usdrl: Unified skeleton-based dense repre- sentation learning with multi-grained feature decorrelation

    Wanjiang Weng, Hongsong Wang, Junbo He, Lei He, and Guosen Xie. Usdrl: Unified skeleton-based dense repre- sentation learning with multi-grained feature decorrelation. arXiv preprint arXiv:2412.09220, 2024. 3, 6, 7

    2024

  27. [35]

    Skeletonmae: Spatial-temporal masked au- toencoders for self-supervised skeleton action recognition

    Wenhan Wu, Yilei Hua, Ce Zheng, Shiqian Wu, Chen Chen, and Aidong Lu. Skeletonmae: Spatial-temporal masked au- toencoders for self-supervised skeleton action recognition. In2023 IEEE International Conference on Multimedia and Expo Workshops (ICMEW), pages 224–229. IEEE, 2023. 3, 6, 7

    2023

  28. [36]

    Deep image clustering with contrastive learn- ing and multi-scale graph convolutional networks.Pattern Recognition, 2024

    Yuankun Xu, Dong Huang, Chang-Dong Wang, and Jian- Huang Lai. Deep image clustering with contrastive learn- ing and multi-scale graph convolutional networks.Pattern Recognition, 2024. 1

    2024

  29. [37]

    Spatial tempo- ral graph convolutional networks for skeleton-based action recognition

    Sijie Yan, Yuanjun Xiong, and Dahua Lin. Spatial tempo- ral graph convolutional networks for skeleton-based action recognition. InThirty-second AAAI Conference on Artificial Intelligence, 2018. 1

    2018

  30. [38]

    View- invariant skeleton action representation learning via motion retargeting.International Journal of Computer Vision, pages 1–16, 2024

    Di Yang, Yaohui Wang, Antitza Dantcheva, Lorenzo Garat- toni, Gianpiero Francesca, and Franc ¸ois Br ´emond. View- invariant skeleton action representation learning via motion retargeting.International Journal of Computer Vision, pages 1–16, 2024. 3

    2024

  31. [39]

    Skeleton cloud colorization for unsupervised 3d action representation learning

    Siyuan Yang, Jun Liu, Shijian Lu, Meng Hwa Er, and Alex C Kot. Skeleton cloud colorization for unsupervised 3d action representation learning. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 13423– 13433, 2021. 3

    2021

  32. [40]

    Self-supervised 3d action representa- tion learning with skeleton cloud colorization.IEEE Trans- actions on Pattern Analysis and Machine Intelligence, 2024

    Siyuan Yang, Jun Liu, Shijian Lu, Er Meng Hwa, Yongjian Hu, and Alex C Kot. Self-supervised 3d action representa- tion learning with skeleton cloud colorization.IEEE Trans- actions on Pattern Analysis and Machine Intelligence, 2024. 1

    2024

  33. [41]

    Contrastive positive mining for unsupervised 3d action representation learning

    Haoyuan Zhang, Yonghong Hou, Wenjing Zhang, and Wan- qing Li. Contrastive positive mining for unsupervised 3d action representation learning. InEuropean Conference on Computer Vision, pages 36–51. Springer, 2022. 2, 3, 6

    2022

  34. [42]

    Prompted con- trast with masked motion modeling: Towards versatile 3d ac- tion representation learning

    Jiahang Zhang, Lilang Lin, and Jiaying Liu. Prompted con- trast with masked motion modeling: Towards versatile 3d ac- tion representation learning. InProceedings of the 31st ACM International Conference on Multimedia, pages 7175–7183,

  35. [43]

    Unsupervised representation learning with long-term dynamics for skeleton based action recognition

    Nenggan Zheng, Jun Wen, Risheng Liu, Liangqu Long, Jian- hua Dai, and Zhefeng Gong. Unsupervised representation learning with long-term dynamics for skeleton based action recognition. InProceedings of the AAAI Conference on Ar- tificial Intelligence, pages 2644–2651, 2018. 1, ...

    2018

Pith tools

Reviewed June 27, 2026 · model on record in the stance chip above.