Pith. sign in

REVIEW 4 major objections 5 minor 46 references

Unsupervised Domain Adaptation for Occlusion Resilient Human Pose Estimation

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read OR-POSE adapts pose estimators to occluded, unlabeled images and reports roughly 7 percent higher keypoint accuracy than prior UDA methods.

desk verdict A plausible and clearly written UDA recipe for occluded pose estimation, with consistent gains over strong baselines, but the '~7%' overstates the weakest result and the batch-relative visibility curriculum needs direct validation. read the letter →

arxiv 2501.02773 v1 pith:MGPF7ZIR submitted 2025-01-06 cs.CV

Arindam Dutta , Sarosij Bose , Saketh Bachu , Calvin-Khang Ta , Konstantinos Karydis , Amit K. Roy-Chowdhury This is my paper · ORCID
classification cs.CV
keywords unsuperviseddomainadaptationhumanposeestimationocclusionresiliencemeanteachercurriculumlearningpriorpseudo-labelingPCK@0.05
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

OR-POSE is an unsupervised domain adaptation method for 2D human pose estimation aimed at the common real-world case where people in unlabeled target images are partially occluded. The paper argues that occlusion can be handled without annotations by training on relatively visible samples first and gradually shifting to heavily occluded ones, using silhouette-derived visibility scores. Across three occluded benchmarks, OR-POSE reports average PCK@0.05 gains of 5.6–7.8 points over the previous state of the art, while staying on par on an unoccluded benchmark, supporting the claim that occlusion resilience and domain adaptation can be combined.

What carries the argument

The load-bearing identity is the visibility score $v_i = \mathrm{sum}(s_i)/\max(\mathrm{sum}(s_i))$, where $s_i$ is the binary silhouette of the visible human in target image $i$ and the maximum is taken over the current training batch; this score weights the prediction-space consistency loss $L^{\text{vis}}_{\text{pred}}$ so that relatively unoccluded images dominate early in adaptation. The curriculum anneals via $\gamma = \exp(-\text{epoch}/\text{total\_epochs})$ in $L_{\text{vis}} = \gamma L^{\text{vis}}_{\text{pred}} + (1-\gamma)L_{\text{pred}}$, starting from visibility-heavy weighting and converging to uniform weighting. The other two pillars are the mean-teacher EMA update that stabilizes pseudo-labels and the parametric prior $G$ (a zero-level set trained on source-plausible poses and corrupted versions) that penalizes anatomically implausible skeletons through the regularizer $L_{\text{ant}}$.

What would settle it

Take a fixed set of 100 target images with known occlusion levels, partition them into two different batches, and compute the visibility scores; if the relative ordering of two images flips between partitions, the curriculum order is an artifact of batch composition. Then train OR-POSE with a visibility score defined globally (e.g., silhouette pixel count divided by a person's bounding-box area) and check whether PCK@0.05 changes materially.

Watch

Extended reading notes

Core claim

The paper claims that a self-training UDA pipeline built on a mean-teacher, an anatomical pose prior, and a visibility curriculum attains state-of-the-art PCK@0.05 on occluded target domains without any target annotations, outperforming UniFrame by 7.7 points on SURREAL→3DOH50K, 7.8 points on SURREAL→Ocl-H36M, and 5.6 points on SURREAL→Ocl-LSP, while staying within 2 points of UniFrame on unoccluded SURREAL→LSP. The result is presented as evidence that occlusion resilience can be achieved in the unsupervised setting by making the training order depend on estimated visibility rather than by requiring paired, multi-view, or temporally continuous supervision.

Load-bearing premise

The method assumes that a silhouette's foreground pixel count, normalized within each training batch, is a reliable ordering of how occluded each person is, so that higher visibility scores genuinely mean less occlusion.

Editorial extensions

If this is right

  • If OR-POSE works as reported, practitioners can adapt a pose estimator to a new, unlabeled, occlusion-heavy deployment setting without collecting keypoint annotations.
  • The 5.6–7.8 point PCK@0.05 improvements over UniFrame across three occluded benchmarks suggest that the combination of anatomical regularization and visibility curricula directly attacks the pseudo-label noise produced by occluded samples.
  • On unoccluded targets, OR-POSE's performance stays within about 2 points of UniFrame, so occlusion resilience does not come at the cost of clean-domain accuracy.
  • The severity experiments show OR-POSE degrades more gracefully than UniFrame as occlusion size increases from roughly 48×48 to 96×96 pixels, indicating robustness across occlusion scales.
  • The ablation attributes specific gains to each component: occlusion augmentation plus pseudo-label consistency yields about 3% over UniFrame, adding the pose prior reaches about 5%, and the visibility curriculum reaches about 7%.

Reading between the lines

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

  • Because the visibility score is normalized by the maximum within a batch, the same target image receives different curriculum weights depending on what else is in the batch; a global calibration of visibility (e.g., using absolute silhouette coverage) would likely make the curriculum more stable and is a natural testable variant.
  • The pose prior is trained only on source-plausible skeletons plus corrupted versions, so its zero-level set may over-penalize valid but unusual target poses; evaluating on a target domain with contorted but legal poses would test whether the prior is too restrictive.
  • The silhouette parser is an off-the-shelf component; if it mislabels occluders as body parts, the visibility signal is corrupted. An experiment with ground-truth occlusion masks replacing the parser would isolate how much of the gain depends on parser quality.
  • The ablation attributes gains to individual components, but components interact; a full factorial ablation would reveal whether the curriculum and the prior are additive or synergistic.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The manuscript proposes OR-POSE, an unsupervised domain adaptation (UDA) method for 2D human pose estimation under occlusion. It combines mean-teacher pseudo-label self-training (Eqs. 2-4), a learned human pose prior used as an anatomical regularizer (Eqs. 5-8), and a visibility-based curriculum loss (Eqs. 9-11). Experiments are reported on SURREAL to 3DOH50K, Ocl-H36M, Ocl-LSP, and non-occluded LSP, comparing against RegDA and UniFrame, with an ablation on SURREAL to 3DOH50K. The central claim is that OR-POSE outperforms analogous state-of-the-art UDA pose estimators by roughly 7% on occluded target datasets while staying on par on non-occluded datasets.

Significance. If confirmed, this is a practically useful contribution: it provides a recipe for adapting a pose estimator to an unlabeled, occlusion-heavy target domain without extra supervision, using standard building blocks (mean teacher, an MLP prior, and a curriculum signal). The benchmark construction with artificially occluded H36M and LSP is sensible, and the ablation in Table V isolates the individual components. The main reservations are that the visibility curriculum is defined batch-relative and is not validated as an occl. , and no error bars or code are provided. Because the headline margin is built on differences of 5-8 PCK points, these issues are load-bearing.

major comments (4)
  1. [Section III-B, Eq. (2)] The teacher update is printed as θ_t^tea ← -α θ_{t-1}^tea + (1-α) θ_t^stu. This is not an exponential moving average; the negative sign means the teacher is not a smoothed version of the student and can oscillate or scale down even when the student converges. If the experiments used the standard EMA update, the equation is a typo and must be corrected; if the printed rule was actually used, the description of the mean-teacher mechanism does not match the method. The paper should state explicitly which rule was used in the experiments.
  2. [Section III-D, Eq. (9)] The visibility score v_i = sum(s_i)/max(sum(s_i)) is normalized by the maximum within each training batch, so the same image receives different weights depending on which other images are in the same batch. It is not an absolute measure of occlusion, and it is confounded by person scale and by errors of the off-the-shelf silhouette model. The paper's claim that the model 'gradually transitions from relatively less occluded to more heavily occluded samples' is not verified: Eq. (11) schedules γ only by epoch, and no experiment shows that Eq. (9) actually orders target images by occlusion severity. This matters because Table V attributes +2.7 PCK on SURREAL to 3DOH50K to adding Lvis (74.2 to 76.9). Please report the correlation between v_i and ground-truth occlusion severity (available for the synthetic Ocl datasets), or replace the batch-relative normalization with a global/absolute score, and provide a sensitivity analysis over batch composition and size.
  3. [Section III-C, Eq. (7)] The training target d for the pose prior is not fully specified. The text says d = 0 for plausible poses and d > 0 for implausible poses, but Eq. (7) minimizes ||G(θ_y) - d||^2; if d is an arbitrary constant for all implausible poses, the prior learns a binary separation and does not provide the distance-to-manifold interpretation claimed in the text. The paper should specify how d is computed for each generated implausible pose. In addition, the implausible poses are generated from source images that are artificially occluded, and no evidence is given that these are representative of the errors the adapted model actually makes on target images. Since Table V attributes +1.7 PCK to Lant (72.5 to 74.2), this is not a purely cosmetic issue.
  4. [Section IV.C, Tables I-IV and VI] All results are reported as single PCK values with no standard deviation, number of runs, or statistical significance testing. The claims in the abstract and Section IV.C that OR-POSE outperforms existing algorithms by ~7% across the three occluded benchmarks are not uniformly supported by the tables: the average gains over UniFrame are 7.7, 7.8, and 5.6 points on SURREAL to 3DOH50K, Ocl-H36M, and Ocl-LSP, respectively. Moreover, Table IV shows OR-POSE at 80.0 vs UniFrame at 82.0 on non-occluded LSP, which does not support the 'on par' claim without error bars. Please report mean and standard deviation over at least three seeds and revise the claim wording to match the per-benchmark margins.
minor comments (5)
  1. [Table V] The fourth row uses the notation '✓ - ✓ ✓' with a dash for Lpred; this is confusing and should be replaced with an explicit indicator, preferably with a footnote explaining that Lpred is absorbed into Lvis via Eq. (11).
  2. [Eq. (10)] The normalization denominator is written as '|B| × P vi', which is ambiguous; it should be written as |B| × Σ_i v_i.
  3. [Section III-D] The paper says binary silhouettes are extracted using pretrained models 'such as [36], [37]', but it does not state which model was actually used, nor how segmentation failures were handled. Please specify the model and the preprocessing.
  4. [Section IV.B] No sensitivity analysis is reported for the hyperparameters τ, λ_a, λ_v, and α; at least a brief study or a reference justifying the chosen values would strengthen the reproducibility of the method.
  5. [Section IV.C] The phrase 'consistently outperforms existing algorithms by ≈7% across all three benchmarks' overstates the Ocl-LSP result (5.6 points); please rephrase to give the actual margins.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity; the central performance claim is an empirical result on external benchmarks, and the only self-citation with evaluative weight is not load-bearing.

full rationale

OR-POSE's derivation chain is not circular: the losses in Eqs. 3, 4, 8, 10-12 are built from source labels, unlabeled target images, and a pose prior trained on source and occluded-source data, while the reported PCK@0.05 numbers are computed against held-out target ground truth. No fitted parameter is renamed as a prediction: the visibility score vi = sum(si)/max(sum(si)) in Eq. 9 is a batch-relative weighting proxy, and the ablation's +2.7 PCK attribution to Lvis is an empirical contribution, not a quantity defined by the evaluation metric. The only self-citations with any role in the evaluation are [14] in related-work context and [42] used for generating Ocl-H36M/Ocl-LSP. Because the main benchmark SURREAL to 3DOH50K is an independent public dataset and the synthetic datasets use external ground-truth keypoints plus Pascal VOC occlusion objects, the headline ~7% margin does not reduce to the authors' own prior results. The batch-relative nature of Eq. 9 is a robustness/correctness concern rather than a circularity, as no target metric or ground-truth label is used to construct the curriculum.

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

The central claim rests on hand-chosen hyperparameters (tau, lambda_a, lambda_v, alpha) and on assumptions about the segmentation mask quality and pose prior transfer. No new physical entities are introduced. The batch-relative visibility score is ad hoc but is a design choice rather than a fitted constant.

free parameters (4)
  • confidence threshold tau = 0.5
    Set based on UniFrame [11]; used to filter pseudo-labels in Eqns. 3 and 10.
  • anatomy loss weight lambda_a = 1e-5
    Chosen by hand and fixed across experiments; weights the pose-prior regularizer in Eqn. 12.
  • visibility loss weight lambda_v = 1
    Chosen by hand; weights Lvis in Eqn. 12.
  • EMA smoothing alpha = not reported
    Eqn. 2 defines teacher updates via alpha, but the paper never states the value used in experiments.
assumptions (4)
  • domain assumption The silhouette segmentation model S produces accurate binary masks of visible human regions on target images.
    Visibility scores in Eqn. 9 depend entirely on these masks; the paper specifies no validation of mask quality.
  • domain assumption The pose prior G trained on source-derived plausible/implausible poses generalizes to the target domain.
    G is trained once on source data (Sec. III-C) and then regularizes all target predictions; no target-specific calibration is performed.
  • ad hoc to paper Batch-relative visibility normalization is a valid curriculum signal.
    Eqn. 9 normalizes by the maximum silhouette sum within each batch, an ad hoc design choice not justified by any analysis.
  • ad hoc to paper The implausible poses used to train G are representative of errors made during adaptation.
    Negatives come from the source model FS applied to occluded source images and von Mises noise [19]; these may not resemble target-domain errors.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unsupervised Domain Adaptation for Occlusion Resilient Human Pose Estimation." pith.science (2026). https://pith.science/paper/MGPF7ZIR

@misc{pith2026250102773,
  author       = {Pith},
  title        = {Pith review of: Unsupervised Domain Adaptation for Occlusion Resilient Human Pose Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MGPF7ZIR}},
  note         = {Machine review of arXiv:2501.02773}
}
abstract

Occlusions are a significant challenge to human pose estimation algorithms, often resulting in inaccurate and anatomically implausible poses. Although current occlusion-robust human pose estimation algorithms exhibit impressive performance on existing datasets, their success is largely attributed to supervised training and the availability of additional information, such as multiple views or temporal continuity. Furthermore, these algorithms typically suffer from performance degradation under distribution shifts. While existing domain adaptive human pose estimation algorithms address this bottleneck, they tend to perform suboptimally when the target domain images are occluded, a common occurrence in real-life scenarios. To address these challenges, we propose OR-POSE: Unsupervised Domain Adaptation for Occlusion Resilient Human POSE Estimation. OR-POSE is an innovative unsupervised domain adaptation algorithm which effectively mitigates domain shifts and overcomes occlusion challenges by employing the mean teacher framework for iterative pseudo-label refinement. Additionally, OR-POSE reinforces realistic pose prediction by leveraging a learned human pose prior which incorporates the anatomical constraints of humans in the adaptation process. Lastly, OR-POSE avoids overfitting to inaccurate pseudo labels generated from heavily occluded images by employing a novel visibility-based curriculum learning approach. This enables the model to gradually transition from training samples with relatively less occlusion to more challenging, heavily occluded samples. Extensive experiments show that OR-POSE outperforms existing analogous state-of-the-art algorithms by $\sim$ 7% on challenging occluded human pose estimation datasets.

Figures

Figures reproduced from arXiv: 2501.02773 by the authors.

Figure 1
Figure 1. Need for unsupervised domain adaptation for occlu￾sion resilient human pose estimation. Left: Predictions of the model trained exclusively on labeled source data (SURREAL) and evaluated on an image from 3DOH50K dataset (referred to as Source only predictions). Middle: Predictions from state￾of-the-art domain adaptive human pose estimation algorithm UniFrame [11]. Right: Predictions from our proposed occlu￾sion resil… view at source ↗
Figure 2
Figure 2. Problem Setup. We propose OR-POSE, an unsu￾pervised algorithm for progressively adapting a model to occlusions. OR-POSE leverages pseudo labels from the mean￾teacher framework to provide guidance to the model while utilizing a pose prior to generate physically plausible poses for humans. Further, to prevent the early-overfitting to noisy pseudo labels and accounting for the uneven levels of occlu￾sion present in the… view at source ↗
Figure 3
Figure 3. Overview of proposed methodology: OR-POSE is built upon the mean-teacher framework, where the weights of the teacher model are updated as an exponential moving average (EMA) of the student model’s weights. OR-POSE uses occlusion augmentations on the source domain, enabling the student and teacher models to provide better pseudo labels on unlabeled target images by learning consistency between occluded and unoccluded… view at source ↗
Figures from the paper (2 more)
Figure 6
Figure 6. Figure 6: Qualitative Results for SURREAL→ Ocl-H36M. From left to right: Source only predictions, prediction from UniFrame [11], and predictions our proposed algo￾rithm OR-POSE [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Qualitative Results for SURREAL→ LSP. From left to right: Source only predictions, prediction from UniFrame [11], and predictions our proposed algo￾rithm OR-POSE. Locl src Lpred Lant Lvis Avg. ✗ ✗ ✗ ✗ 57.1 ✓ ✓ ✗ ✗ 72.5 ✓ ✓ ✓ ✗ 74.2 ✓ - ✓ ✓ 76.9 TABLE V: Ablation study …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 36 canonical work pages

  1. [1]

    Pgvt: Pose-guided video transformer for fine-grained action recognition,

    H. Zhang, M. C. Leong, L. Li, and W. Lin, “Pgvt: Pose-guided video transformer for fine-grained action recognition,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2024, pp. 6645–6656

    work page 2024

  2. [2]

    Seas: Shape- aligned supervision for person re-identification,

    H. Zhu, P. Budhwant, Z. Zheng, and R. Nevatia, “Seas: Shape- aligned supervision for person re-identification,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 164–174

    work page 2024

  3. [3]

    Gaitgraph: Graph convolutional network for skeleton-based gait recog- nition,

    T. Teepe, A. Khan, J. Gilg, F. Herzog, S. H ¨ormann, and G. Rigoll, “Gaitgraph: Graph convolutional network for skeleton-based gait recog- nition,” in 2021 IEEE International Conference on Image Processing (ICIP). IEEE, 2021, pp. 2314–2318

    work page 2021

  4. [4]

    Higherhrnet: Scale-aware representation learning for bottom-up human pose estimation,

    B. Cheng, B. Xiao, J. Wang, H. Shi, T. S. Huang, and L. Zhang, “Higherhrnet: Scale-aware representation learning for bottom-up human pose estimation,” in CVPR, 2020

    work page 2020

  5. [5]

    Zoomnas: searching for whole-body human pose estimation in the wild,

    L. Xu, S. Jin, W. Liu, C. Qian, W. Ouyang, P. Luo, and X. Wang, “Zoomnas: searching for whole-body human pose estimation in the wild,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 4, pp. 5296–5313, 2022

    work page 2022

  6. [6]

    Effective whole-body pose estimation with two-stages distillation,

    Z. Yang, A. Zeng, C. Yuan, and Y . Li, “Effective whole-body pose estimation with two-stages distillation,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 4210–4220

    work page 2023

  7. [7]

    Human3.6m: Large scale datasets and predictive methods for 3d human sensing in natural environments,

    C. Ionescu, D. Papava, V . Olaru, and C. Sminchisescu, “Human3.6m: Large scale datasets and predictive methods for 3d human sensing in natural environments,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 36, no. 7, pp. 1325–1339, jul 2014

    2014

  8. [8]

    Microsoft coco: Common objects in context,

    T.-Y . Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Doll ´ar, and C. L. Zitnick, “Microsoft coco: Common objects in context,” in Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 13 . Springer, 2014, pp. 740–755

    2014

Show all 46 references
  1. [9]

    Being’seen’vs.’mis-seen’: Tensions between privacy and fairness in computer vision,

    A. Xiang, “Being’seen’vs.’mis-seen’: Tensions between privacy and fairness in computer vision,” Harvard Journal of Law & Technology , vol. 36, no. 1, 2022

    2022

  2. [10]

    Regressive domain adaptation for unsupervised keypoint detection,

    J. Jiang, Y . Ji, X. Wang, Y . Liu, J. Wang, and M. Long, “Regressive domain adaptation for unsupervised keypoint detection,” in CVPR, 2021

    2021

  3. [11]

    A unified framework for domain adaptive pose estimation,

    D. Kim, K. Wang, K. Saenko, M. Betke, and S. Sclaroff, “A unified framework for domain adaptive pose estimation,” in ECCV, 2022

    2022

  4. [12]

    Occlusion-aware siamese network for human pose estimation,

    L. Zhou, Y . Chen, Y . Gao, J. Wang, and H. Lu, “Occlusion-aware siamese network for human pose estimation,” inComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XX 16 . Springer, 2020, pp. 396–412

    2020

  5. [13]

    Source-free domain adaptive human pose estimation,

    Q. Peng, C. Zheng, and C. Chen, “Source-free domain adaptive human pose estimation,” in Proceedings of the IEEE/CVF International Con- ference on Computer Vision , 2023, pp. 4826–4836

    2023

  6. [14]

    Prior-guided source-free domain adaptation for human pose estimation,

    D. S. Raychaudhuri, C.-K. Ta, A. Dutta, R. Lal, and A. K. Roy- Chowdhury, “Prior-guided source-free domain adaptation for human pose estimation,” in Proceedings of the IEEE/CVF International Con- ference on Computer Vision , 2023, pp. 14 996–15 006

    2023

  7. [15]

    3d human pose estimation using spatio-temporal networks with explicit occlusion training,

    Y . Cheng, B. Yang, B. Wang, and R. T. Tan, “3d human pose estimation using spatio-temporal networks with explicit occlusion training,” in Proceedings of the AAAI Conference on Artificial Intelligence , vol. 34, no. 07, 2020, pp. 10 631–10 638

    2020

  8. [16]

    Explicit occlusion reasoning for multi-person 3d human pose estimation,

    Q. Liu, Y . Zhang, S. Bai, and A. Yuille, “Explicit occlusion reasoning for multi-person 3d human pose estimation,” in European Conference on Computer Vision . Springer, 2022, pp. 497–517

    2022

  9. [17]

    Mean teachers are better role mod- els: Weight-averaged consistency targets improve semi-supervised deep learning results,

    A. Tarvainen and H. Valpola, “Mean teachers are better role mod- els: Weight-averaged consistency targets improve semi-supervised deep learning results,” NeurIPS, 2017

    2017

  10. [18]

    How robust is 3d human pose estimation to occlusion?

    I. Sarandi, T. Linder, K. O. Arras, and B. Leibe, “How robust is 3d human pose estimation to occlusion?” in 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) . IEEE, 2020, pp. 10 302–10 309

    2020

  11. [19]

    Pose-ndf: Modeling human pose manifolds with neural distance fields,

    G. Tiwari, D. Antic, J. E. Lenssen, N. Sarafianos, T. Tung, and G. Pons- Moll, “Pose-ndf: Modeling human pose manifolds with neural distance fields,” in ECCV, 2022

    2022

  12. [20]

    C-sfda: A curriculum learning aided self-training framework for efficient source free domain adaptation,

    N. Karim, N. C. Mithun, A. Rajvanshi, H.-p. Chiu, S. Samarasekera, and N. Rahnavard, “C-sfda: A curriculum learning aided self-training framework for efficient source free domain adaptation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tio...

    2023

  13. [21]

    Deep learning based 2d human pose estimation: A survey,

    Q. Dang, J. Yin, B. Wang, and W. Zheng, “Deep learning based 2d human pose estimation: A survey,” Tsinghua Science and Technology , vol. 24, no. 6, pp. 663–676, 2019

    2019

  14. [22]

    Rmpe: Regional multi-person pose estimation,

    H.-S. Fang, S. Xie, Y .-W. Tai, and C. Lu, “Rmpe: Regional multi-person pose estimation,” in ICCV, 2017

    2017

  15. [23]

    Mask r-cnn,

    K. He, G. Gkioxari, P. Doll ´ar, and R. Girshick, “Mask r-cnn,” in ICCV, 2017. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 9

    2017

  16. [24]

    Simple baselines for human pose estimation and tracking,

    B. Xiao, H. Wu, and Y . Wei, “Simple baselines for human pose estimation and tracking,” in ECCV, 2018

    2018

  17. [25]

    Deep high-resolution represen- tation learning for human pose estimation,

    K. Sun, B. Xiao, D. Liu, and J. Wang, “Deep high-resolution represen- tation learning for human pose estimation,” in CVPR, 2019

    2019

  18. [26]

    Openpose: realtime multi-person 2d pose estimation using part affinity fields,

    Z. Cao, G. Hidalgo, T. Simon, S.-E. Wei, and Y . Sheikh, “Openpose: realtime multi-person 2d pose estimation using part affinity fields,” IEEE transactions on Pattern Analysis and Machine Intelligence , vol. 43, no. 1, pp. 172–186, 2019

    2019

  19. [27]

    Differentiable hierarchical graph grouping for multi-person pose esti- mation,

    S. Jin, W. Liu, E. Xie, W. Wang, C. Qian, W. Ouyang, and P. Luo, “Differentiable hierarchical graph grouping for multi-person pose esti- mation,” in ECCV, 2020

    2020

  20. [28]

    Bottom-up human pose estimation via disentangled keypoint regression,

    Z. Geng, K. Sun, B. Xiao, Z. Zhang, and J. Wang, “Bottom-up human pose estimation via disentangled keypoint regression,” arXiv preprint arXiv:2104.02300, 2021

    2021 arXiv

  21. [29]

    Peeking into occluded joints: A novel framework for crowd pose estimation,

    L. Qiu, X. Zhang, Y . Li, G. Li, X. Wu, Z. Xiong, X. Han, and S. Cui, “Peeking into occluded joints: A novel framework for crowd pose estimation,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XIX

    2020

  22. [30]

    Springer, 2020, pp. 488–504

    2020

  23. [31]

    Ocr- pose: Occlusion-aware contrastive representation for unsupervised 3d human pose estimation,

    J. Wang, Z. Yu, Z. Tong, H. Wang, J. Liu, W. Zhang, and X. Wu, “Ocr- pose: Occlusion-aware contrastive representation for unsupervised 3d human pose estimation,” in Proceedings of the 30th ACM International Conference on Multimedia , 2022, pp. 5477–5485

    2022

  24. [32]

    Cross- domain adaptation for animal pose estimation,

    J. Cao, H. Tang, H.-S. Fang, X. Shen, C. Lu, and Y .-W. Tai, “Cross- domain adaptation for animal pose estimation,” in Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 9498– 9507

    2019

  25. [33]

    Maps: A noise- robust progressive learning approach for source-free domain adaptive keypoint detection,

    Y . Ding, J. Liang, B. Jiang, A. Zheng, and R. He, “Maps: A noise- robust progressive learning approach for source-free domain adaptive keypoint detection,” IEEE Transactions on Circuits and Systems for Video Technology, 2023

    2023

  26. [34]

    Deeppose: Human pose estimation via deep neural networks,

    A. Toshev and C. Szegedy, “Deeppose: Human pose estimation via deep neural networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2014, pp. 1653–1660

    2014

  27. [35]

    Leap: Learning articulated occupancy of people,

    M. Mihajlovic, Y . Zhang, M. J. Black, and S. Tang, “Leap: Learning articulated occupancy of people,” in CVPR, 2021

    2021

  28. [36]

    The generalized von mises distribution,

    R. Gatto and S. R. Jammalamadaka, “The generalized von mises distribution,” Statistical Methodology, vol. 4, no. 3, pp. 341–353, 2007

    2007

  29. [37]

    Self-correction for human pars- ing,

    P. Li, Y . Xu, Y . Wei, and Y . Yang, “Self-correction for human pars- ing,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 44, no. 6, pp. 3260–3271, 2020

    2020

  30. [38]

    Rethinking atrous convolution for semantic image segmentation,

    L.-C. Chen, G. Papandreou, F. Schroff, and H. Adam, “Rethinking atrous convolution for semantic image segmentation,” arXiv preprint arXiv:1706.05587, 2017

    2017 arXiv

  31. [39]

    Learning from synthetic humans,

    G. Varol, J. Romero, X. Martin, N. Mahmood, M. J. Black, I. Laptev, and C. Schmid, “Learning from synthetic humans,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2017, pp. 109–117

    2017

  32. [40]

    Object-occluded human shape and pose estimation from a single color image,

    T. Zhang, B. Huang, and Y . Wang, “Object-occluded human shape and pose estimation from a single color image,” in IEEE Conference on Computer Vision and Pattern Recognition, (CVPR) , 2020

    2020

  33. [41]

    Clustered pose and nonlinear ap- pearance models for human pose estimation

    S. Johnson and M. Everingham, “Clustered pose and nonlinear ap- pearance models for human pose estimation.” in bmvc, vol. 2, no. 4. Aberystwyth, UK, 2010, p. 5

    2010

  34. [42]

    Human de- occlusion: Invisible perception and recovery for humans,

    Q. Zhou, S. Wang, Y . Wang, Z. Huang, and X. Wang, “Human de- occlusion: Invisible perception and recovery for humans,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion, 2021, pp. 3691–3701

    2021

  35. [43]

    Poise: Pose guided human silhouette extraction under occlusions,

    A. Dutta, R. Lal, D. S. Raychaudhuri, C.-K. Ta, and A. K. Roy- Chowdhury, “Poise: Pose guided human silhouette extraction under occlusions,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2024, pp. 6153–6163

    2024

  36. [44]

    The pascal visual object classes (voc) challenge,

    M. Everingham, L. Van Gool, C. K. Williams, J. Winn, and A. Zisser- man, “The pascal visual object classes (voc) challenge,” International journal of computer vision , vol. 88, pp. 303–338, 2010

    2010

  37. [45]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 770–778

    2016

  38. [46]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980 , 2014

    2014 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.