Pith. sign in

REVIEW 3 major objections 6 minor 53 references

Multiview-Consistent Semi-Supervised Learning for 3D Human Pose Estimation

T0 review · 3 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read MCSS learns a multi-view-consistent pose embedding from unlabeled synchronized video and regresses a canonical, yaw-invariant 3D pose from it with limited 3D labels, improving normalized pose error from 149.28 mm to 111.94 mm with S1-only…

desk verdict Solid weak-supervision method for 3D pose with a nice multi-view embedding idea, but the headline comparison to Rhodin et al. rests on an unspecified evaluation frame and a canonicalization equation that is ambiguous as written. read the letter →

arxiv 1908.05293 v3 pith:RVU32RYQ submitted 2019-08-14 cs.CV

Rahul Mitra , Nitesh B. Gundavarapu , Abhishek Sharma , Arjun Jain This is my paper · ORCID
classification cs.CV
keywords 3dhumanposeestimationsemi-supervisedlearningmulti-viewconsistencymetrichardnegativeminingcanonicalretrievallimitedsupervision
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 proposes MCSS, a semi-supervised framework that converts cheap, unlabeled synchronized multi-view video into a pose-specific learning signal for monocular 3D human pose estimation. It trains a shared embedding so that different viewpoints of the same intrinsic pose land close together while different poses separate, using a contrastive loss with in-batch hard-negative mining. With 3D labels from only one training subject, this signal cuts normalized per-joint error from 149.28 mm (supervised-only baseline) to 111.94 mm and beats the leading geometry-aware alternative's 122.60 mm, all on a smaller ResNet-18 backbone and without 2D pose labels, camera extrinsics, or background extraction. The paper also introduces view-invariant pose retrieval as a benchmark on two public multi-view pose datasets.

What carries the argument

The central object is a 128-dimensional pose embedding $\varphi$ shared by the contrastive loss and the pose-regression head. The contrastive loss (Eqs. 1-2) pulls same-pose, different-view pairs together and pushes the closest in-batch negatives apart via a margin $\alpha$ and threshold $\beta$; mini-batches drawn from one subject with overlapping backgrounds stop the embedding from using appearance or background to separate negatives. The canonical pose transformation (Eq. 3) rotates each MoCap skeleton about the vertical (+Z) axis by angle $\theta = \cos^{-1}(\hat{u}_{xy}\cdot \hat{\imath})$ so the pelvis-to-left-hip bone is parallel to the XZ plane, giving a view- and rigid-rotation-invariant regression target. Because the embedding already enforces consistency, a shallow FC(128,48) head suffices and the backbone is only ResNet-18, unlike the ResNet-50 used by the comparison method [36].

What would settle it

Take a MoCap skeleton whose pelvis-to-left-hip direction has a negative y-component and apply Eq. 3 as written; if the resulting canonical skeleton is a mirrored version of the one produced by its x-axis reflection, the transformation is not the bijection the argument needs. A public implementation check, whether the rotation uses a signed angle and whether an inverse transform to camera coordinates is defined, would settle both the representation and the reported MPJPE comparisons.

Watch

Extended reading notes

Core claim

The paper claims that a multi-view-consistent pose embedding, learned from synchronized multi-view video by contrastive metric learning, is enough weak supervision to train a monocular 3D pose regressor, provided the regression target is a canonical pose that removes viewpoint and rigid-rotation ambiguity. The embedding is shared between the contrastive loss and a shallow regression head, so limited 3D pose labels refine the embedding while the embedding constrains pose predictions. With labels from a single subject, MCSS reaches 111.94 mm N-MPJPE against 149.28 mm for the supervised baseline and 122.60 mm for the leading geometry-aware method [36]; the improvement persists when that method is restricted to the same ResNet-18 backbone. The same embedding also supports cross-view, cross-subject pose retrieval with errors close to an oracle that has ground-truth 3D poses.

Load-bearing premise

The load-bearing premise is that Eq. 3 defines a consistent bijection from every global MoCap pose to a canonical pose: as written, $\theta = \arccos(\hat{u}_{xy}\cdot \hat{\imath})$ ignores the sign of the y-component, so symmetric poses may not reach a unique canonical target, and the paper never states how canonical predictions are mapped back to camera coordinates for MPJPE.

Editorial extensions

If this is right

  • With one subject's 3D labels on the main benchmark, MCSS reports 111.94 mm N-MPJPE, versus 149.28 mm for the supervised-only baseline and 122.60 mm for the prior geometry-aware method [36].
  • The method degrades gracefully as 3D supervision shrinks from all five subjects to 5% of one subject, showing the contrastive signal carries pose structure.
  • Regressing canonical pose instead of global-coordinate pose is worth about 45 N-MPJPE points in their ablation (MCSS-global 157.30 vs MCSS 111.94).
  • The learned embedding retrieves cross-view, cross-subject poses with Mean-PA-MPJPE@K close to the ground-truth oracle on both multi-view datasets, supporting pose retrieval as a usable downstream task.
  • No 2D pose labels, pre-trained 2D pose estimators, camera extrinsics, or background extraction are required, and inference is 24.8 ms per image versus 75.3 ms for the comparison model [36].

Reading between the lines

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

  • A direct test of the mechanism would be to train MCSS on multi-view footage with near-identical camera angles; if the contrastive signal still helps, view consistency is not what drives the gain.
  • The cross-dataset drop the paper reports (retrieval MPJPE 119.6 mm without fine-tuning) suggests the embedding retains domain-specific appearance cues; a domain-adversarial variant might separate pose content from appearance more cleanly.
  • Eq. 3 as written uses arccos of a dot product that ignores the sign of the y-component, so mirrored poses could map inconsistently unless the implementation uses a signed angle; making the rotation bijective and specifying the inverse transform would let MPJPE comparisons to camera-coordinate methods be reproduced.
  • The same embedding could serve as a differentiable pose prior for single-image 3D reconstruction in settings where only multi-view video, not 3D labels, is available; the paper's retrieval experiments are evidence the descriptor is pose-specific.
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

3 major / 6 minor

Summary. The paper proposes MCSS, a semi-supervised framework for monocular 3D human pose estimation. It trains a ResNet-18 feature extractor and an embedding network with a contrastive metric-learning loss on synchronized, unlabeled multi-view video frames, using hard-negative mining, while simultaneously regressing a 'canonical' 3D pose from the embedding with limited 3D labels. The canonical representation is defined in Eq. (3) by rotating the global MoCap pose about the vertical axis so that the pelvis-to-left-hip bone lies parallel to the XZ plane. Experiments on Human3.6M and MPI-INF-3DHP report that under S1-only supervision MCSS reaches N-MPJPE 111.94 mm versus 149.28 mm for the baseline and 122.60 mm for Rhodin et al. [36], and the paper also proposes cross-view pose-retrieval benchmarks.

Significance. If the reported numbers are measured on a common footing, the paper makes a useful contribution: it shows that synchronized, uncalibrated multi-view video can supply a metric-learning signal that reduces dependence on 3D pose labels, and it does so with a smaller backbone (ResNet-18) than [36] (ResNet-50). The canonical-pose target is a sensible response to the one-to-many ambiguity between the learned embedding and view-specific coordinates, and the global-versus-canonical ablation in Table 2 is informative. The proposed cross-view and cross-subject pose-retrieval protocol, with Oracle comparisons, is a useful benchmark for future work. However, the central quantitative claim hinges on coordinate-frame details that are not specified; until those are resolved, the improvement over [36] cannot be taken at face value.

major comments (3)
  1. [Section 3.2.1, Eq. (3)] The canonicalization is not a well-defined function of the pose as written. Because theta = cos^{-1}(u_xy dot i) returns the absolute angle rather than a signed angle, a left-hip horizontal projection with a positive y component is rotated the wrong way, and the resulting bone is not parallel to the XZ plane. The same intrinsic pose can therefore receive different canonical targets depending on which side of the XZ plane it lies, making the L1 pose-regression supervision inconsistent. The paper should define theta with a signed angle, e.g., theta = -atan2(u_y, u_x), and state explicitly that the implementation uses the signed version; no code is provided to disambiguate. In addition, the inverse of Eq. (3) is never defined, although it is needed if any evaluation is done after mapping predictions back to global coordinates.
  2. [Section 5, Table 1 and Fig. 3] The evaluation frame for MPJPE, N-MPJPE, and PA-MPJPE is never stated. MCSS outputs canonical poses, whereas Rhodin et al. [36,37] output camera-coordinate poses. If the metrics are computed in the canonical frame, the comparison is not on a common footing, because canonicalization applies a pose-dependent rotation R(theta(p)): the quantity ||R(theta(p_hat)) p_hat - R(theta(p_gt)) p_gt|| is not equal to ||p_hat - p_gt|| in general, and part of the reported gap could be a coordinate artifact. If the metrics are computed after transforming canonical predictions back to global or camera coordinates, the inverse of Eq. (3) must be specified. Please state, for each metric, which coordinate frame is used, how [36] and the baseline are placed in that same frame, and whether the canonical rotation is computed from the prediction, the ground truth, or a shared value. Without this information, the headline improvement of 25% over baseline and 8.7% over state-of-the-art cannot be verified.
  3. [Section 5, Table 2] The ablation against MCSS-global is difficult to interpret without the evaluation protocol. MCSS-global is trained to regress poses in the MoCap global coordinate system; if it is then evaluated under a canonical-frame metric, its substantially larger error could be partly an artifact of comparing a non-canonical output against a canonical target, rather than evidence about the learned representation. Please specify how MCSS-global's predictions are evaluated, for example whether they are canonicalized at test time or measured directly in global coordinates, and report the metric in a frame that treats both variants fairly.
minor comments (6)
  1. [Section 5, paragraph after Fig. 3] The text uses 'MVSS' where the method is elsewhere called 'MCSS'; please correct this typo.
  2. [Section 4, Pose Regression paragraph] The text refers to 'The pose regression network G', but Fig. 1a and Sec. 3.2 denote the regression head as H; please align the notation.
  3. [Section 3.1, Eq. (1)] If all candidate distances in the batch are below beta, the product delta(D(i,j)) D(i,j) is zero for every j, so jmin and kmin are not uniquely defined; please state how this degenerate case is handled.
  4. [References] References [6] and [7] are the same paper; one duplicate should be removed.
  5. [Section 6.3, Generalization] The sentence reporting 119.6 mm for K=10 and 101.9 mm for K=20 appears to contradict the intuition and the trend in Table 3; please clarify whether K is the number of retrieved neighbors and why a larger K gives lower error in this cross-dataset setting.
  6. [Section 3.1.1 and Section 4] The paper states that specific hyper-parameter choices are detailed in the supplementary material; if the supplementary material is part of the submission, it should be included with the review version.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the semi-supervised signal and pose regression are measured against external benchmarks and independent baselines.

full rationale

I walked the derivation chain and found no step where a claimed prediction reduces by construction to an input or to a self-citation. The semi-supervised signal Lcnstr (Eq. 2) is built from synchronized multiview image pairs and hard-negative mining, which is independent of the 3D pose labels used in Lpose (Sec. 3.2). Pose estimation is evaluated on held-out subjects S9/S11 against an internal baseline and external methods (Rhodin et al. [36] and [37]), with results reported on public datasets Human3.6M and MPI-INF-3DHP. References [8] and [40] are self-citations from the same group, but they appear only in general context and related-work statements; no load-bearing claim, uniqueness theorem, or ansatz is imported from them. The canonical pose representation (Eq. 3) is defined directly from MoCap global coordinates, and the paper explicitly ablates it against global-coordinate regression (MCSS vs. MCSS-global, Table 2), so the representation choice is tested rather than assumed. The ambiguity about how canonical predictions are mapped back for MPJPE comparison is an evaluation-protocol concern and a correctness risk, not a circularity. The derivation is self-contained relative to external benchmarks.

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

The method's extra signal comes from a synchronization assumption, a hand-set negative-mining heuristic, and a canonical pose target. The central comparisons are against external methods on public benchmarks. No new physical entities are introduced.

free parameters (4)
  • Contrastive margin alpha = 0.6
    Hand-set margin in Eq. 2 controlling separation of hard negatives; final results depend on this value, but no sensitivity analysis or sweep is reported.
  • Hard-negative threshold beta = 0.3
    Hand-set minimum distance in Eq. 1 that determines which in-batch samples count as negatives; critical to avoid treating similar poses as negatives, and no ablation is reported.
  • Embedding dimension dim_phi = 128
    Dimension of the learned embedding in Sec. 4; chosen by hand following descriptor-learning literature, with no ablation on this choice.
  • Metric/pose batch ratio = 3:1
    Training ratio between the metric-learning and pose-regression batches in Sec. 4; hand-set and central to joint training, but no sensitivity analysis is provided.
assumptions (4)
  • domain assumption Frames captured synchronously from different views of the same subject contain the same intrinsic 3D pose.
    Anchor-positive construction in Sec. 3.1 relies on synchronized multi-view video labels without 3D annotations; if synchronization or pose equivalence fails, the contrastive positives are mislabeled and the weak supervision signal is corrupted.
  • domain assumption Within a same-subject mini-batch, the closest non-positive samples are valid hard negatives of different poses, and the threshold beta prevents pushing apart temporally close same-pose frames.
    Mini-batch selection in Sec. 3.1.1 and Eq. 1 assume same-subject appearance and background cannot separate negatives, so the embedding must use pose; the choice beta=0.3 is a heuristic with no ablation or sensitivity analysis.
  • ad hoc to paper Eq. 3 defines a well-defined yaw-invariant canonicalization for all global poses, so the same intrinsic pose always maps to the same canonical target.
    Canonical pose is introduced in Sec. 3.2.1 to resolve the one-to-many embedding-to-pose mapping. As written, theta = cos^-1(uxy dot i) drops the sign of the y component, so for roughly half the hip directions the rotated bone is not parallel to the XZ plane; the assumption is therefore not guaranteed by the stated math.
  • domain assumption ImageNet-pretrained ResNet features transfer to multi-view pose similarity and 3D pose regression.
    Sec. 4 initializes all models from ImageNet weights; the interaction between pretraining and the semi-supervised signal is never isolated, so part of the claimed gain could come from the pretrained backbone.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multiview-Consistent Semi-Supervised Learning for 3D Human Pose Estimation." pith.science (2026). https://pith.science/paper/RVU32RYQ

@misc{pith2026190805293,
  author       = {Pith},
  title        = {Pith review of: Multiview-Consistent Semi-Supervised Learning for 3D Human Pose Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RVU32RYQ}},
  note         = {Machine review of arXiv:1908.05293}
}
read the original abstract

The best performing methods for 3D human pose estimation from monocular images require large amounts of in-the-wild 2D and controlled 3D pose annotated datasets which are costly and require sophisticated systems to acquire. To reduce this annotation dependency, we propose Multiview-Consistent Semi Supervised Learning (MCSS) framework that utilizes similarity in pose information from unannotated, uncalibrated but synchronized multi-view videos of human motions as additional weak supervision signal to guide 3D human pose regression. Our framework applies hard-negative mining based on temporal relations in multi-view videos to arrive at a multi-view consistent pose embedding. When jointly trained with limited 3D pose annotations, our approach improves the baseline by 25% and state-of-the-art by 8.7%, whilst using substantially smaller networks. Lastly, but importantly, we demonstrate the advantages of the learned embedding and establish view-invariant pose retrieval benchmarks on two popular, publicly available multi-view human pose datasets, Human 3.6M and MPI-INF-3DHP, to facilitate future research.

Figures

Figures reproduced from arXiv: 1908.05293 by the authors.

Figure 1
Figure 1. (a) Training framework for learning our pose embedding and subsequent canonical pose estimation. xanchor and xpositive are a batch of anchor and positive image pairs taken from different camera views. xsupervise is the batch of images with 3D-pose supervision. F is the ResNet based feature extractor. G maps features extracted from F to our embedding φ. The Hard Negative Sampling module performs in-batch hard mining … view at source ↗
Figure 2
Figure 2. Shows top view of bottom half of a human skele [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. N-MPJPE vs. 3D-pose supervision on test split of Human3.6M. Our proposed model outperforms the baseline and the current state-of-the-art Rhodin et al. [36]. In Tab. 1, we compare MPJPE, N-MPJPE and PA￾MPJPE values of our approach against baseline and [36]. Clearly, our method outperforms [36] by 22.4 N-MPJPE when fully supervised on 3D data and by 10.7 N-MPJPE with 3D-pose supervision limited to S1. For MPJPE how￾ev… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Qualitative results on canonical pose estimation by our proposed framework (MCSS) against our Baseline on Human 3.6M test split (S9, S11). Both the models are trained with supervision from labels of subject S1. Our method produces more accurate estimates for even for c…
Figure 5
Figure 5. Figure 5: Qualitative image retrievals on Human 3.6M (S9, [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Mean-PA-MPJPE for increasing number of re [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Variation of mean embedding distance with in [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 48 canonical work pages

  1. [36]

    Unsu- pervised geometry-aware representation for 3d human pose estimation

    Helge Rhodin, Mathieu Salzmann, and Pascal Fua. Unsu- pervised geometry-aware representation for 3d human pose estimation. In ECCV, 2018. 2, 3, 5, 6

    work page 2018

  2. [1]

    Densepose: Dense human pose estimation in the wild

    Rıza Alp G ¨uler, Natalia Neverova, and Iasonas Kokkinos. Densepose: Dense human pose estimation in the wild. In CVPR, 2018. 3

    2018

  3. [2]

    2d human pose estimation: New benchmark and state of the art analysis

    Mykhaylo Andriluka, Leonid Pishchulin, Peter Gehler, and Bernt Schiele. 2d human pose estimation: New benchmark and state of the art analysis. In CVPR, 2014. 5, 6

    work page 2014

  4. [3]

    Ex- ploiting temporal context for 3d human pose estimation in the wild

    Anurag Arnab, Carl Doersch, and Andrew Zisserman. Ex- ploiting temporal context for 3d human pose estimation in the wild. In CVPR, 2019. 3

    work page 2019

  5. [4]

    Position referencing and consistent world modeling for mobile robots

    Raja Chatila and Jean-Paul Laumond. Position referencing and consistent world modeling for mobile robots. In ICRA,

    work page

  6. [5]

    Unsupervised 3d pose estimation with geometric self- supervision

    Ching-Hang Chen, Ambrish Tyagi, Amit Agrawal, Dy- lan Drover, Rohith MV , Stefan Stojanov, and James M Rehg. Unsupervised 3d pose estimation with geometric self- supervision. In CVPR, 2019. 3, 5

    work page 2019

  7. [7]

    Weakly-supervised discovery of geometry-aware representation for 3d human pose estimation

    Xipeng Chen, Kwan-Yee Lin, Wentao Liu, Chen Qian, and Liang Lin. Weakly-supervised discovery of geometry-aware representation for 3d human pose estimation. In CVPR,

    work page

  8. [8]

    Learning 3d hu- man pose from structure and motion

    Rishabh Dabral, Anurag Mundhada, Uday Kusupati, Safeer Afaque, Abhishek Sharma, and Arjun Jain. Learning 3d hu- man pose from structure and motion. In ECCV, 2018. 1, 3

    work page 2018

Show all 53 references
  1. [9]

    Inferring 3d body pose from silhouettes using activity manifold learning

    Ahmed Elgammal and Chan-Su Lee. Inferring 3d body pose from silhouettes using activity manifold learning. In CVPR,

  2. [10]

    Semi- supervised learning in gigantic image collections

    Rob Fergus, Yair Weiss, and Antonio Torralba. Semi- supervised learning in gigantic image collections. In NIPS,

  3. [11]

    Holopose: Holistic 3d human reconstruction in-the-wild

    Riza Alp Guler and Iasonas Kokkinos. Holopose: Holistic 3d human reconstruction in-the-wild. In CVPR, 2019. 3

    2019

  4. [12]

    In the wild human pose estimation using explicit 2d features and intermediate 3d rep- resentations

    Ikhsanul Habibie, Weipeng Xu, Dushyant Mehta, Gerard Pons-Moll, and Christian Theobalt. In the wild human pose estimation using explicit 2d features and intermediate 3d rep- resentations. In CVPR, 2019. 3

    2019

  5. [13]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR,

  6. [14]

    Catalin Ionescu, Dragos Papava, Vlad Olaru, and Cristian Sminchisescu. Human3. 6m: Large scale datasets and pre- dictive methods for 3d human sensing in natural environ- ments. In TPAMI, 2013. 1, 5

    2013

  7. [15]

    End-to-end recovery of human shape and pose

    Angjoo Kanazawa, Michael J Black, David W Jacobs, and Jitendra Malik. End-to-end recovery of human shape and pose. In CVPR, 2018. 3

    2018

  8. [16]

    Learning 3d human dynamics from video

    Angjoo Kanazawa, Jason Y Zhang, Panna Felsen, and Jiten- dra Malik. Learning 3d human dynamics from video. In CVPR, 2019. 3

    2019

  9. [17]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. In arXiv, 2014. 5

    2014

  10. [18]

    Self- supervised learning of 3d human pose using multi-view ge- ometry

    Muhammed Kocabas, Salih Karagoz, and Emre Akbas. Self- supervised learning of 3d human pose using multi-view ge- ometry. In CVPR, 2019. 2, 3, 5

    2019

  11. [19]

    Affine structure from motion

    Jan J Koenderink and Andrea J Van Doorn. Affine structure from motion. In JOSA A, 1991. 1

    1991

  12. [20]

    Learning to reconstruct 3d human pose and shape via model-fitting in the loop

    Nikos Kolotouros, Georgios Pavlakos, Michael J Black, and Kostas Daniilidis. Learning to reconstruct 3d human pose and shape via model-fitting in the loop. In CVPR, 2019. 3

    2019

  13. [21]

    Semi- supervised deep learning for monocular depth map predic- tion

    Yevhen Kuznietsov, Jorg Stuckler, and Bastian Leibe. Semi- supervised deep learning for monocular depth map predic- tion. In CVPR, 2017. 1

    2017

  14. [22]

    Thin-slicing for pose: Learning to understand pose without explicit pose es- timation

    Suha Kwak, Minsu Cho, and Ivan Laptev. Thin-slicing for pose: Learning to understand pose without explicit pose es- timation. In CVPR, 2016. 2, 7

    2016

  15. [23]

    Modeling view and posture manifolds for tracking

    Chan-Su Lee and Ahmed Elgammal. Modeling view and posture manifolds for tracking. In CVPR, 2007. 2

    2007

  16. [24]

    Semi-supervised boosting using visual similarity learning

    Christian Leistner, Helmut Grabner, and Horst Bischof. Semi-supervised boosting using visual similarity learning. In CVPR, 2008. 1

    2008

  17. [25]

    Generating multiple hypotheses for 3d human pose estimation with mixture density network

    Chen Li and Gim Hee Lee. Generating multiple hypotheses for 3d human pose estimation with mixture density network. In CVPR, 2019. 3

    2019

  18. [26]

    Visualizing data using t-sne

    Laurens van der Maaten and Geoffrey Hinton. Visualizing data using t-sne. In JMLR, 2008. 6

    2008

  19. [27]

    A simple yet effective baseline for 3d human pose estimation

    Julieta Martinez, Rayat Hossain, Javier Romero, and James J Little. A simple yet effective baseline for 3d human pose estimation. In CVPR, 2017. 3, 4

    2017

  20. [28]

    Monocular 3d human pose estimation in the wild using improved cnn supervision

    Dushyant Mehta, Helge Rhodin, Dan Casas, Pascal Fua, Oleksandr Sotnychenko, Weipeng Xu, and Christian Theobalt. Monocular 3d human pose estimation in the wild using improved cnn supervision. In 3DV, 2017. 1, 5

    2017

  21. [29]

    Vnect: Real-time 3d human pose estimation with a single rgb cam- era

    Dushyant Mehta, Srinath Sridhar, Oleksandr Sotnychenko, Helge Rhodin, Mohammad Shafiei, Hans-Peter Seidel, Weipeng Xu, Dan Casas, and Christian Theobalt. Vnect: Real-time 3d human pose estimation with a single rgb cam- era. In TOG, 2017. 1, 3

    2017

  22. [30]

    Working hard to know your neighbor’s mar- gins: Local descriptor learning loss

    Anastasiia Mishchuk, Dmytro Mishkin, Filip Radenovic, and Jiri Matas. Working hard to know your neighbor’s mar- gins: Local descriptor learning loss. In NIPS, 2017. 3, 5

    2017

  23. [31]

    A survey of com- puter vision-based human motion capture

    Thomas B Moeslund and Erik Granum. A survey of com- puter vision-based human motion capture. In CVIU, 2001. 1

    2001

  24. [32]

    Pose embeddings: A deep architecture for learning to match human poses

    Greg Mori, Caroline Pantofaru, Nisarg Kothari, Thomas Le- ung, George Toderici, Alexander Toshev, and Weilong Yang. Pose embeddings: A deep architecture for learning to match human poses. In arXiv, 2015. 2

    2015

  25. [33]

    Coarse-to-fine volumetric predic- tion for single-image 3d human pose

    Georgios Pavlakos, Xiaowei Zhou, Konstantinos G Derpa- nis, and Kostas Daniilidis. Coarse-to-fine volumetric predic- tion for single-image 3d human pose. In CVPR, 2017. 1, 4

    2017

  26. [34]

    3d human pose estimation in video with tem- poral convolutions and semi-supervised training

    Dario Pavllo, Christoph Feichtenhofer, David Grangier, and Michael Auli. 3d human pose estimation in video with tem- poral convolutions and semi-supervised training. In CVPR,

  27. [35]

    Deep multitask architecture for integrated 2d and 3d human sensing

    Alin-Ionut Popa, Mihai Zanfir, and Cristian Sminchisescu. Deep multitask architecture for integrated 2d and 3d human sensing. In CVPR, 2017. 3 9

    2017

  28. [37]

    Learning monocular 3d human pose esti- mation from multi-view images

    Helge Rhodin, J ¨org Sp ¨orri, Isinsu Katircioglu, Victor Con- stantin, Fr ´ed´eric Meyer, Erich M ¨uller, Mathieu Salzmann, and Pascal Fua. Learning monocular 3d human pose esti- mation from multi-view images. In CVPR, 2018. 3, 4, 5, 6

    2018

  29. [38]

    Imagenet large scale visual recognition challenge

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, San- jeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, et al. Imagenet large scale visual recognition challenge. In IJCV, 2015. 5

    2015

  30. [39]

    A comparison and eval- uation of multi-view stereo reconstruction algorithms

    Steven M Seitz, Brian Curless, James Diebel, Daniel Scharstein, and Richard Szeliski. A comparison and eval- uation of multi-view stereo reconstruction algorithms. In CVPR, 2006. 1

    2006

  31. [40]

    Monocular 3d human pose estimation by generation and ordinal ranking

    Saurabh Sharma, Pavan Teja Varigonda, Prashast Bindal, Abhishek Sharma, and Arjun Jain. Monocular 3d human pose estimation by generation and ordinal ranking. In ICCV,

  32. [41]

    Semi-supervised learning for anomalous trajectory detection

    Rowland R Sillito and Robert B Fisher. Semi-supervised learning for anomalous trajectory detection. InBMVC, 2008. 1

    2008

  33. [42]

    Self- supervised learning of pose embeddings from spatiotemporal relations in videos

    Omer Sumer, Tobias Dencker, and Bjorn Ommer. Self- supervised learning of pose embeddings from spatiotemporal relations in videos. In ICCV, 2017. 2, 7

    2017

  34. [43]

    Compositional human pose regression

    Xiao Sun, Jiaxiang Shang, Shuang Liang, and Yichen Wei. Compositional human pose regression. In ICCV, 2017. 1, 3

    2017

  35. [44]

    Integral human pose regression

    Xiao Sun, Bin Xiao, Fangyin Wei, Shuang Liang, and Yichen Wei. Integral human pose regression. In ECCV, 2018. 1, 3, 4, 5

    2018

  36. [45]

    Structured prediction of 3d human pose with deep neural networks

    Bugra Tekin, Isinsu Katircioglu, Mathieu Salzmann, Vincent Lepetit, and Pascal Fua. Structured prediction of 3d human pose with deep neural networks. In BMVC, 2016. 3

    2016

  37. [46]

    L2-net: Deep learn- ing of discriminative patch descriptor in euclidean space

    Yurun Tian, Bin Fan, and Fuchao Wu. L2-net: Deep learn- ing of discriminative patch descriptor in euclidean space. In CVPR, 2017. 5

    2017

  38. [47]

    Lifting from the deep: Convolutional 3d pose estimation from a sin- gle image

    Denis Tome, Chris Russell, and Lourdes Agapito. Lifting from the deep: Convolutional 3d pose estimation from a sin- gle image. In CVPR, 2017. 1, 3, 4, 5

    2017

  39. [48]

    3d people tracking with gaussian process dynamical models

    Raquel Urtasun, David J Fleet, and Pascal Fua. 3d people tracking with gaussian process dynamical models. In CVPR,

  40. [49]

    3d hu- man pose estimation with siamese equivariant embedding

    M ´arton V ´eges, Viktor Varga, and Andr ´as L ˝orincz. 3d hu- man pose estimation with siamese equivariant embedding. In Neurocomputing, 2019. 3

    2019

  41. [50]

    Repnet: Weakly su- pervised training of an adversarial reprojection network for 3d human pose estimation

    Bastian Wandt and Bodo Rosenhahn. Repnet: Weakly su- pervised training of an adversarial reprojection network for 3d human pose estimation. In CVPR, 2019. 3, 4

    2019

  42. [51]

    3d human pose machines with self-supervised learning

    Keze Wang, Liang Lin, Chenhan Jiang, Chen Qian, and Pengxu Wei. 3d human pose machines with self-supervised learning. In TPAMI, 2019. 3

    2019

  43. [52]

    Drpose3d: Depth ranking in 3d hu- man pose estimation

    Min Wang, Xipeng Chen, Wentao Liu, Chen Qian, Liang Lin, and Lizhuang Ma. Drpose3d: Depth ranking in 3d hu- man pose estimation. In IJCAI, 2018. 1

    2018

  44. [53]

    Semantic graph convolutional networks for 3d human pose regression

    Long Zhao, Xi Peng, Yu Tian, Mubbasir Kapadia, and Dim- itris N Metaxas. Semantic graph convolutional networks for 3d human pose regression. In CVPR, 2019. 3

    2019

  45. [54]

    Weakly-supervised transfer for 3d human pose estimation in the wild

    Xingyi Zhou, Qixing Huang, Xiao Sun, Xiangyang Xue, and Yichen Wei. Weakly-supervised transfer for 3d human pose estimation in the wild. In ICCV, 2017. 1, 3, 4 10

    2017

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.