Pith. sign in

REVIEW 4 major objections 4 minor 3 cited by

USDRL: Unified Skeleton-Based Dense Representation Learning with Multi-Grained Feature Decorrelation

T0 review · 4 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read USDRL claims that a multi-grained feature-decorrelation loss on pooled temporal, spatial, and instance vectors learns dense skeleton representations that outperform previous self-supervised methods on recognition, retrieval, and detection.

desk verdict Solid empirical self-supervised skeleton method with strong results, but the 'dense representation' claim rests on a pooled loss rather than per-frame supervision and needs a from-scratch baseline to hold up. read the letter →

arxiv 2412.09220 v2 pith:DZ6YXJNT submitted 2024-12-12 cs.CV

Wanjiang Weng , Hongsong Wang , Junbo Wang , Lei He , Guosen Xie This is my paper · ORCID
classification cs.CV
keywords skeleton-basedactionrecognitionself-supervisedrepresentationlearningfeaturedecorrelationdensedetectionretrievalspatio-temporaltransformernegative-sample-free
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

USDRL is a self-supervised method that learns skeleton representations by decorrelating features—making different dimensions of the representation statistically independent—instead of contrasting positive and negative pairs. The paper claims this single multi-grained objective, applied to temporal, spatial, and instance domains, produces dense, frame-level representations that work for recognition, retrieval, and detection alike. On NTU-60, NTU-120, PKU-MMD I, and PKU-MMD II, it reports accuracies and detection mAPs above previous self-supervised methods, with the largest gap in action detection. If the claim holds, skeleton pretraining can drop the momentum encoder, memory bank, and negative-pair engineering while improving downstream dense tasks.

What carries the argument

The engine is the Multi-Grained Feature Decorrelation loss ($L_{f d}$, Eq. 7). It combines an intra-sample consistency term (mean-squared-error similarity plus an invariance term that pulls autocorrelations toward the identity) with an inter-sample separability term built from variance, auto-covariance, and cross-correlation penalties. These losses are computed in three projected spaces—instance, temporal, and spatial—on condensed vectors obtained by max-pooling the dense outputs $y_t$ and $y_s$. The dense outputs themselves come from the Dense Spatio-Temporal Encoder (DSTE), which stacks two modules per layer: Dense Shift Attention (DSA), an MLP plus masked token-mixing operation called DenseShift followed by self-attention and feed-forward blocks, and Convolutional Attention (CA), a 1D convolution before self-attention. A weighted sum of the two branches yields the temporal and spatial dense representations that carry the frame-level information.

What would settle it

Use the released USDRL checkpoint and evaluate frame-level action detection on PKU-MMD I twice: once with the dense per-frame outputs $y_t$ and once with the pooled vector broadcast across frames. If per-frame and pooled versions give similar mAP, the per-frame features are not carrying the claimed signal; if per-frame is much better, the dense representation is doing real work.

Watch

Extended reading notes

Core claim

The paper's central claim is that feature decorrelation is not only a collapse-prevention trick but a complete learning signal for skeleton representations, and that it can work at several granularities at once. The training objective pushes projected vectors from the same sequence to be close and their autocorrelations to look like the identity, while variance, auto-covariance, and cross-correlation terms force different dimensions and different samples to be independent. Because the same loss is applied in temporal, spatial, and instance domains, the encoder is encouraged to keep information along both axes rather than collapsing a sequence into a single global vector. On the benchmarks, USDRL with the new DSTE encoder reports top-1 recognition of 85.2% on NTU-60 x-sub (87.1% with a three-modality ensemble), retrieval gains of about 1 to 4 points over the best compared methods, and action-detection mAP of 75.7/74.9 on PKU-MMD I versus 61.8/61.3 for the strongest previously published method. The takeaway, if accepted, is that dense skeleton representations can be learned without any negative samples.

Load-bearing premise

The load-bearing premise is that a decorrelation loss applied only to max-pooled, projected vectors transfers to the per-frame outputs of the encoder; if the frame-level features are not actually shaped by this loss, the dense-representation claim collapses even if recognition results remain valid.

Editorial extensions

If this is right

  • If the reported results hold, self-supervised skeleton pretraining no longer needs a memory bank, momentum encoder, or tuned negative-pair sampling; the same pipeline serves recognition, retrieval, and detection.
  • Multi-grained decorrelation transfers across backbones: replacing DSTE with the standard STTR still improves over previous methods, so the loss, not only the encoder, carries much of the benefit.
  • For action detection, the choice of dense encoder matters: DSTE outperforms STTR within the same framework by about 9 points of mAP on PKU-MMD I, suggesting frame-level tasks are where the architecture's dense outputs pay off.
  • Single-modality joint training is competitive with multi-modal ensembling: USDRL with joints alone beats UmURL using joints plus motion plus bone on most recognition protocols, implying modality engineering may be less critical than the objective.

Reading between the lines

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

  • Not tested in the paper: because the published loss touches only pooled vectors, adding an explicit per-frame decorrelation term to Eq. 7 would be the sharpest test of whether the dense claim is real, and could push detection accuracy further.
  • If the pooled-to-frame transfer holds, the framework should scale to longer sequences and larger skeleton datasets without the memory-bank cost of contrastive methods; the paper does not run this scaling experiment.
  • The DenseShift mask's gap controls how much global context each token absorbs; varying it could expose a tradeoff between global recognition and local detection accuracy, which the paper does not explore.
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. The paper proposes USDRL, a self-supervised skeleton-based representation learning method that avoids negative samples by using feature decorrelation losses inspired by Barlow Twins and VICReg. The framework operates on temporal, spatial, and instance domains, and introduces a Dense Spatio-Temporal Encoder (DSTE) composed of Convolutional Attention and Dense Shift Attention modules. The authors evaluate the method on NTU-60, NTU-120, PKU-MMD I, and PKU-MMD II for action recognition, retrieval, and detection, reporting consistent improvements over prior self-supervised methods, especially on action detection.

Significance. If the reported results hold, USDRL would show that feature decorrelation alone can match or exceed negative-based contrastive methods in skeleton-based representation learning, while also providing a route to dense (frame-level) features for action detection. The paper includes a broad set of benchmarks, ablations, and code availability, which are strengths. The main concern is that the central claim of learning 'dense representations' is not directly supported by the pretraining objective, which is applied only to global pooled vectors; this needs additional evidence or rephrasing.

major comments (4)
  1. [Multi-Grained Feature Decorrelation, Eq. (8)] Equation (8) contains a dimensional inconsistency: bz_a and bz_b are described as normalized vectors, so bz_a^T bz_b is a scalar, while I is an identity matrix; the expression tr(I - bz_a^T bz_b) is therefore not well-defined. If an autocorrelation matrix is intended (e.g., bz_a bz_b^T), this must be stated explicitly. This ambiguity prevents the reader from reproducing the exact loss used in training.
  2. [Multi-Grained Feature Decorrelation, Eq. (11) and surrounding text] The text explains the cross-correlation term by stating that off-diagonal elements 'derived from two different samples' should approach zero. However, in Eq. (11), the cross-correlation matrix Xcorr(Za, Zb) is indexed by feature dimensions (i, j), so off-diagonal entries correspond to correlations between different feature dimensions, not between different samples. This misdescription misrepresents the mechanism of the loss and should be corrected.
  3. [Methods / Overall training objective] The paper's central claim is that USDRL 'learns dense representations through multi-grained feature decorrelation.' However, the total loss in Eq. (7) is computed solely on the projected vectors zt, zs, and zi, which are obtained after MaxPooling the dense representations yt and ys (Methods). No term in Eqs. (8)-(13) directly supervises the per-frame or per-joint entries of yt and ys. Consequently, the pretraining objective is a global, sequence-level loss. The action detection results in Table 4 are obtained by fine-tuning the entire DSTE on PKU-MMD I with frame-wise labels, so they do not isolate the contribution of the self-supervised pretraining to dense feature quality. To support the dense representation claim, the authors should add a from-scratch baseline (no pretraining) for detection and/or evaluate frozen dense features (e.g., linear probing at the frame level) without fine-tuning the encoder.
  4. [Experimental setup / Comparison fairness] The paper compares single-modality (joint) USDRL with multi-modality baselines in Table 1 and claims superiority, but the comparison is not always apples-to-apples because the DSTE model is trained with a different architecture and potentially different augmentation schedules. The significance of the improvements should be contextualized by reporting the number of parameters and training cost of DSTE versus STTR and other baselines, and by explaining how the evaluation protocol (e.g., frozen encoder linear evaluation) aligns with previous work.
minor comments (4)
  1. [Table 4 caption] The caption contains a typo: 'Comparsion' should be 'Comparison.'
  2. [Conclusion] The word 'Spaito-Temporal' should be 'Spatio-Temporal.'
  3. [Table 5 header] The header uses 'V AC' but the text refers to the variance/auto-covariance terms; this should be written as 'VAC' or 'V+AC' for consistency.
  4. [Section Multi-Grained Feature Decorrelation] The phrase 'Intra-sample Separately loss' appears to be a typo; it should read 'Inter-sample Separability loss.'

Circularity Check

0 steps flagged · score 2.0 of 10

No derivation reduces to its inputs; the only self-citation is minor and not load-bearing.

full rationale

The paper's derivation chain is not circular. The multi-grained decorrelation objective (Eqs. 7-13) is explicitly assembled from Barlow Twins (Zbontar et al. 2021) and VICREG (Bardes, Ponce, and LeCun 2022) terms, which are external, independently published objectives, and the encoder is evaluated by standard linear probing (frozen feature extractor plus trained classifier) and by full fine-tuning with labels. No equation is defined in terms of the reported accuracies, and no fitted parameter is renamed as a prediction: the hyperparameters (tau, mu, lambda, gamma, epsilon, alpha, beta) are not recovered from the test set through the loss equations. The detection experiments fine-tune the entire model with frame-wise labels on PKU-MMD I, so the large gains over STTR in Table 4 are a supervised fine-tuning result rather than a claim forced by the pretraining loss; at most this supports a reviewer concern that the 'dense representation' claim is not directly supervised by Eq. 7, but that is an overclaim or evaluation gap, not a circular reduction. The one overlapping-author citation, Zhou et al. 2023 (J. Wang), is used only to name the feature-decorrelation paradigm; the actual loss formulas are attributed to Zbontar et al. 2021 and Bardes et al. 2022, and no result depends on a uniqueness theorem or an unverified claim from that citation. Therefore no circularity step can be exhibited, and the appropriate finding is no significant circularity; the score of 2 reflects only the minor non-load-bearing self-citation.

Assumptions & free parameters 9 free parameters · 4 assumptions · 3 invented entities

The central claim depends on several hand-chosen hyperparameters, none of which are reported with values, and on the untested assumption that global decorrelation of pooled vectors transfers to dense per-frame features. The DSTE and its modules are new architectural entities supported only by the paper's own experiments.

free parameters (9)
  • tau (τ) = unspecified
    Balances the instance-domain loss against the spatial and temporal losses in Eq. 7; chosen by hand, no value given.
  • kappa (κ) = unspecified
    Weight for the Similarity term in Eq. 8; chosen by hand, no value given.
  • eta (η) = unspecified
    Weight for the Invariance term in Eq. 8; chosen by hand, no value given.
  • mu (μ) = unspecified
    Weight for the Variance term in Eq. 12; chosen by hand, no value given.
  • lambda (λ) = unspecified
    Weight for the Cross-Correlation term in Eq. 12; chosen by hand, no value given.
  • gamma (γ) = unspecified
    Variance threshold in Eq. 9; standard VICREG constant, but the exact value is not reported.
  • alpha (α) = unspecified
    Weight between CA and DSA in Eq. 5; ablated in Fig. 4, but the optimal value is not named in the text.
  • K (number of augmentations) = unspecified
    Used in Eqs. 8 and 12; the paper never states how many augmentations are used.
  • Training schedule (batch size, learning rate, epochs, optimizer) = unspecified
    No training hyperparameters are reported anywhere in the paper.
assumptions (4)
  • domain assumption Max pooling the dense representations preserves enough information for both global and dense downstream tasks.
    All pretraining losses in Eq. 7 operate on pooled vectors; the paper does not justify that this supervises per-frame features.
  • domain assumption Feature decorrelation on augmented skeleton views prevents model collapse without negative samples.
    The mechanism is borrowed from VICREG and Barlow Twins and assumed to transfer to skeleton sequences.
  • domain assumption Splitting the input into temporal and spatial streams captures the action's spatio-temporal structure.
    The two-stream design is asserted in the framework overview and not derived from any analysis.
  • ad hoc to paper The mask-blend operation in Eq. 2 constitutes a 'DenseShift' operation.
    Eq. 2 is a binary mask mixture; its relationship to a shift operation is asserted with a citation to Xing et al. 2023, without a formal definition or derivation.
invented entities (3)
  • Dense Spatio-Temporal Encoder (DSTE)
    purpose: A two-stream encoder intended to produce frame-level dense representations from skeleton sequences.
    Validated only through internal ablations and benchmark results in this paper; no external analysis or formal characterization.
  • Dense Shift Attention (DSA)
    purpose: An attention module that combines MLP-based global modeling with a masked blend to enrich dense dependencies.
    The 'shift' mechanism is not formally defined, and the module is only evaluated as part of the full system.
  • Convolutional Attention (CA)
    purpose: A module combining 1D convolution with self-attention to integrate local features and long-range dependencies.
    Composed of standard components; the specific arrangement has no independent support outside the paper's ablations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of USDRL: Unified Skeleton-Based Dense Representation Learning with Multi-Grained Feature Decorrelation." pith.science (2026). https://pith.science/paper/DZ6YXJNT

@misc{pith2026241209220,
  author       = {Pith},
  title        = {Pith review of: USDRL: Unified Skeleton-Based Dense Representation Learning with Multi-Grained Feature Decorrelation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DZ6YXJNT}},
  note         = {Machine review of arXiv:2412.09220}
}
read the original abstract

Contrastive learning has achieved great success in skeleton-based representation learning recently. However, the prevailing methods are predominantly negative-based, necessitating additional momentum encoder and memory bank to get negative samples, which increases the difficulty of model training. Furthermore, these methods primarily concentrate on learning a global representation for recognition and retrieval tasks, while overlooking the rich and detailed local representations that are crucial for dense prediction tasks. To alleviate these issues, we introduce a Unified Skeleton-based Dense Representation Learning framework based on feature decorrelation, called USDRL, which employs feature decorrelation across temporal, spatial, and instance domains in a multi-grained manner to reduce redundancy among dimensions of the representations to maximize information extraction from features. Additionally, we design a Dense Spatio-Temporal Encoder (DSTE) to capture fine-grained action representations effectively, thereby enhancing the performance of dense prediction tasks. Comprehensive experiments, conducted on the benchmarks NTU-60, NTU-120, PKU-MMD I, and PKU-MMD II, across diverse downstream tasks including action recognition, action retrieval, and action detection, conclusively demonstrate that our approach significantly outperforms the current state-of-the-art (SOTA) approaches. Our code and models are available at https://github.com/wengwanjiang/USDRL.

Figures

Figures reproduced from arXiv: 2412.09220 by the authors.

Figure 1
Figure 1. Illustration of the feature decorrelation-based self [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The proposed Unified Skeleton-based Dense Representation Learning (USDRL) framework. USDRL incorporates [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The basic layer of Dense Spatio-Temporal En [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: The impact of weight hyperparameter α for action recognition on the xsub evaluation of the NTU-60 dataset. Specifically, upon receiving an action query, the nearest neighbor is identified within the representation space us￾ing cosine similarity. Results shown in [PITH…
Figure 5
Figure 5. Figure 5: Visualizations of learned instance-level representations obtained by (a) Negative Contrastive Learning (CL), (b) Multi [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

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

  1. Towards Efficient General Feature Prediction in Masked Skeleton Modeling

    cs.CV 2025-09 conditional novelty 6.0 of 10

    GFP predicts hierarchical high-level features from a jointly trained lightweight target network instead of reconstructing joint coordinates, speeding up masked skeleton pretraining 6.2x while improving downstream accuracy.

  2. Frequency-Guided Diffusion Model with Perturbation Training for Skeleton-Based Video Anomaly Detection

    cs.CV 2024-12 conditional novelty 6.0 of 10

    A frequency-guided diffusion model with perturbation training improves skeleton-based video anomaly detection accuracy on five benchmarks.

  3. DuoCLR: Dual-Surrogate Contrastive Learning for Skeleton-based Human Action Segmentation

    cs.CV 2025-09 conditional novelty 5.0 of 10

    DuoCLR pretrains on trimmed skeleton sequences using Shuffle-and-Warp multi-action permutations and two surrogate tasks, significantly improving action segmentation on untrimmed videos.

Reference graph

Works this paper leans on

45 extracted references · 40 canonical work pages · cited by 3 Pith papers

  1. [1]

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

    work page 2022

  2. [2]

    Chen, Y.; Zhao, L.; Yuan, J.; Tian, Y.; Xia, Z.; Geng, S.; Han, L.; and Metaxas, D. N. 2022. Hierarchically self-supervised transformer for human skeleton representation learning. In European Conference on Computer Vision, 185--202. Springer

    work page 2022

  3. [3]

    Chen, Z.; Wang, H.; and Gui, J. 2023. Occluded Skeleton-Based Human Action Recognition with Dual Inhibition Training. In Proceedings of the 31st ACM International Conference on Multimedia, 2625--2634

    work page 2023

  4. [4]

    H.; Chi, S.; Lee, S

    Chi, H.-g.; Ha, M. H.; Chi, S.; Lee, S. W.; Huang, Q.; and Ramani, K. 2022. InfoGCN: Representation Learning for Human Skeleton-Based Action Recognition. In Proceedings of the IEEE/CVF Conference on Computer vision and Pattern Recognition (CVPR), 20186--20196

    work page 2022

  5. [5]

    Dong, J.; Sun, S.; Liu, Z.; Chen, S.; Liu, B.; and Wang, X. 2023. Hierarchical Contrast for Unsupervised Skeleton-based Action Representation Learning. In Proceedings of the AAAI Conference on Artificial Intelligence

    work page 2023

  6. [6]

    Ermolov, A.; Siarohin, A.; Sangineto, E.; and Sebe, N. 2021. Whitening for self-supervised representation learning. In International conference on machine learning, 3015--3024. PMLR

    work page 2021

  7. [7]

    Franco, L.; Mandica, P.; Munjal, B.; and Galasso, F. 2023. Hyperbolic Self-paced Learning for Self-supervised Skeleton-based Action Representations. In The Eleventh International Conference on Learning Representations

    work page 2023

  8. [8]

    Grill, J.-B.; Strub, F.; Altch \'e , F.; Tallec, C.; Richemond, P.; Buchatskaya, E.; Doersch, C.; Avila Pires, B.; Guo, Z.; Gheshlaghi Azar, M.; et al. 2020. Bootstrap your own latent-a new approach to self-supervised learning. Advances in neural information processing systems, 33: 21271--21284

    2020

Show all 45 references
  1. [9]

    He, K.; Chen, X.; Xie, S.; Li, Y.; Doll \'a r, P.; and Girshick, R. 2022. Masked autoencoders are scalable vision learners. In Proceedings of the IEEE/CVF Conference on Computer vision and Pattern Recognition, 16000--16009

    2022

  2. [10]

    Jing, L.; Vincent, P.; LeCun, Y.; and Tian, Y. 2022. Understanding Dimensional Collapse in Contrastive Self-supervised Learning. In International Conference on Learning Representations

    2022

  3. [11]

    J.; Kim, J.; and Choi, J

    Kim, B.; Chang, H. J.; Kim, J.; and Choi, J. Y. 2022. Global-local motion transformer for unsupervised skeleton-based action learning. In European conference on computer vision, 209--225. Springer

    2022

  4. [12]

    Lin, L.; Song, S.; Yang, W.; and Liu, J. 2020. Ms2l: Multi-task self-supervised learning for skeleton based action recognition. In Proceedings of the 28th ACM international conference on multimedia, 2490--2498

    2020

  5. [13]

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

    2023

  6. [14]

    Linguo, L.; Minsi, W.; Bingbing, N.; Hang, W.; Jiancheng, Y.; and Wenjun, Z. 2021. 3D Human Action Representation Learning via Cross-View Consistency Pursuit. In CVPR

    2021

  7. [15]

    Liu, J.; Shahroudy, A.; Perez, M.; Wang, G.; Duan, L.-Y.; and Kot, A. C. 2019. 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

  8. [16]

    Liu, J.; Song, S.; Liu, C.; Li, Y.; and Hu, Y. 2020. 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

  9. [17]

    Liu, Z.; Gui, J.; and Luo, H. 2023. Good helper is around you: Attention-driven masked image modeling. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, 1799--1807

    2023

  10. [18]

    Mao, Y.; Deng, J.; Zhou, W.; Fang, Y.; Ouyang, W.; and Li, H. 2023. Masked motion predictors are strong 3d action representation learners. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 10181--10191

    2023

  11. [19]

    Mao, Y.; Zhou, W.; Lu, Z.; Deng, J.; and Li, H. 2022. Cmd: Self-supervised 3d action representation learning with cross-modal mutual distillation. In European Conference on Computer Vision, 734--752. Springer

    2022

  12. [20]

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

    2021

  13. [21]

    Shah, A.; Roy, A.; Shah, K.; Mishra, S.; Jacobs, D.; Cherian, A.; and Chellappa, R. 2023. 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 (CVPR), 1...

    2023

  14. [22]

    Shahroudy, A.; Liu, J.; Ng, T.-T.; and Wang, G. 2016. Ntu rgb+ d: A large scale dataset for 3d human activity analysis. In Proceedings of the IEEE Conference on Computer vision and Pattern Recognition, 1010--1019

    2016

  15. [23]

    T.; Xie, E.; Zhou, B.; Xing, L.; Chandrasekaran, A.; et al

    Sun, J.; Huang, L.; Wang, H.; Zheng, C.; Qiu, J.; Islam, M. T.; Xie, E.; Zhou, B.; Xing, L.; Chandrasekaran, A.; et al. 2024. Localization and recognition of human action in 3D using transformers. Communications Engineering, 3(1): 125

    2024

  16. [24]

    Sun, S.; Liu, D.; Dong, J.; Qu, X.; Gao, J.; Yang, X.; Wang, X.; and Wang, M. 2023. Unified Multi-modal Unsupervised Representation Learning for Skeleton-based Action Understanding. In Proceedings of the 31st ACM International Conference on Multimedia, 2973--2984

    2023

  17. [25]

    M.; Doughty, H.; and Snoek, C

    Thoker, F. M.; Doughty, H.; and Snoek, C. G. 2021. Skeleton-contrastive 3D action representation learning. In Proceedings of the 29th ACM international conference on multimedia, 1655--1663

    2021

  18. [26]

    Tianyu, G.; Hong, L.; Zhan, C.; Mengyuan, L.; Tao, W.; and Runwei, D. 2022. Contrastive Learning from Extremely Augmented Skeleton Sequences for Self-supervised Action Recognition. In AAAI

    2022

  19. [27]

    Wang, H.; and Wang, L. 2018. Beyond joints: Learning representations from primitive geometries for skeleton-based action recognition and detection. IEEE Transactions on Image Processing, 27(9): 4382--4394

    2018

  20. [28]

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

    2022

  21. [29]

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

    2024

  22. [30]

    Wu, C.; Wu, X.-J.; Kittler, J.; Xu, T.; Ahmed, S.; Awais, M.; and Feng, Z. 2024. SCD-Net: Spatiotemporal Clues Disentanglement Network for Self-supervised Skeleton-based Action Recognition. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 5949--5957

    2024

  23. [31]

    Xie, Z.; Zhang, Z.; Cao, Y.; Lin, Y.; Bao, J.; Yao, Z.; Dai, Q.; and Hu, H. 2022. Simmim: A simple framework for masked image modeling. In Proceedings of the IEEE/CVF Conference on Computer vision and Pattern Recognition, 9653--9663

    2022

  24. [32]

    Xing, J.; Wang, M.; Ruan, Y.; Chen, B.; Guo, Y.; Mu, B.; Dai, G.; Wang, J.; and Liu, Y. 2023. Boosting few-shot action recognition with graph-guided hybrid matching. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 1740--1750

    2023

  25. [33]

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

    2023

  26. [34]

    Yang, D.; Wang, Y.; Dantcheva, A.; Garattoni, L.; Francesca, G.; and Br \'e mond, F. 2024. View-invariant Skeleton Action Representation Learning via Motion Retargeting. International Journal of Computer Vision, 1--16

    2024

  27. [35]

    M.; Hu, Y.; and Kot, A

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

    2023

  28. [36]

    Yang, Z.; Liang, J.; Xu, Y.; Zhang, X.-Y.; and He, R. 2023 b . Masked relation learning for deepfake detection. IEEE Transactions on Information Forensics and Security, 18: 1696--1708

    2023

  29. [37]

    Zbontar, J.; Jing, L.; Misra, I.; LeCun, Y.; and Deny, S. 2021. Barlow twins: Self-supervised learning via redundancy reduction. In International conference on machine learning, 12310--12320. PMLR

    2021

  30. [38]

    Zhang, H.; Hou, Y.; Zhang, W.; and Li, W. 2022. Contrastive positive mining for unsupervised 3d action representation learning. In European Conference on Computer Vision, 36--51. Springer

    2022

  31. [39]

    Zhang, J.; Lin, L.; and Liu, J. 2023 a . Hierarchical consistent contrastive learning for skeleton-based action recognition with growing augmentations. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, 3427--3435

    2023

  32. [40]

    Zhang, J.; Lin, L.; and Liu, J. 2023 b . Prompted Contrast with Masked Motion Modeling: Towards Versatile 3D Action Representation Learning. In Proceedings of the 31st ACM International Conference on Multimedia, 7175--7183

    2023

  33. [41]

    Zhou, Y.; Duan, H.; Rao, A.; Su, B.; and Wang, J. 2023. Self-supervised action representation learning from partial spatio-temporal skeleton sequences. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, 3825--3833

    2023

  34. [42]

    Zhu, W.; Ma, X.; Liu, Z.; Liu, L.; Wu, W.; and Wang, Y. 2023 a . Motionbert: A unified perspective on learning human motion representations. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 15085--15099

    2023

  35. [43]

    Zhu, Y.; Han, H.; Yu, Z.; and Liu, G. 2023 b . Modeling the relative visual tempo for self-supervised skeleton-based action recognition. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 13913--13922

    2023

  36. [44]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  37. [45]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.