Pith. sign in

REVIEW 4 major objections 4 minor 39 references

VirtualFencer: Generating Fencing Bouts based on Strategies Extracted from In-the-Wild Videos

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read VirtualFencer recovers 3D fencing motion and right-of-way strategy from unlabeled broadcast video, and professional saber fencers rate its self-play touches as indistinguishable from ground-truth touches.

desk verdict A solid systems paper with real expert evaluation, but the core strategy-extraction claim rests on an unvalidated displacement-based priority heuristic. read the letter →

arxiv 2507.00261 v1 pith:5VRU2M6U submitted 2025-06-30 cs.CV cs.GR

Zhiyin Lin , Purvi Goel , Joy Yun , C. Karen Liu , Joao Pedro Araujo This is my paper · ORCID
classification cs.CVcs.GR
keywords fencingunsupervisedskilldiscoverymotionreconstructionstrategymodelingrightofwayself-playuserstudybroadcastvideo
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to show that a complete tactical model of fencing can be built without any annotated motion-capture data, using only unlabeled broadcast video. The system reconstructs both fencers' 3D motion, discovers a vocabulary of low-level skills through clustering, and learns how those skills are sequenced under different right-of-way states. The central evidence is a user study in which over 20 professional saber fencers could not distinguish touches generated by the system's self-play from real touches, while clearly preferring both over random baselines. If correct, this means realistic, strategic sports simulation can be scaled to any sport with abundant broadcast footage, without expensive manual annotation.

What carries the argument

The central machinery is a three-part pipeline: (1) a homography-corrected pose-estimation stage that maps both fencers' global positions onto the canonical 14-meter piste, using WHAM for SMPL pose and SAM2-prompted tracking of piste lines; (2) a two-stage $k$-means clustering over 20-frame subclips embedded with TMR features, dominant-arm axis-angle joints, and distance statistics, yielding 30 labeled skill clusters; and (3) a priority-aware strategy model: three separate first-order transition matrices conditioned on the priority state (M-M, P-NP, NP-P), combined with a Gaussian distance weight at inference time, and a simulation state machine that updates priority by scoring lights, finishing actions, or displacement fallback, and terminates on out-of-bounds, crash, touch, or terminal action.

What would settle it

Take a sample of touches from the 40 collected bouts, compare the heuristic priority labels from Section 5.3 against the official referee's right-of-way decisions (recoverable from the scoring lights and the bout outcome), and measure the disagreement rate; a high disagreement rate around exchanges would show the strategy model is conditioned on noisy labels. A stronger test would re-run the self-play user study with referee-verified priority labels instead of the heuristic and check whether the indistinguishability result persists.

Watch

Extended reading notes

Core claim

The authors claim that a fully unsupervised pipeline from monocular broadcast video to a strategy-conditioned generative model of fencing touches works: they extract about 1.5 hours of motion from 40 senior-level bouts, correct the global translations of the off-the-shelf pose estimator using a homography aligned to the fencing piste, cluster motion embeddings into 30 discrete skill clusters, and fit three priority-mode transition distributions of the form $P(u_t | u_{t-1}, v_{t-1}, d_t)$ with a Gaussian distance weighting. The learned model is then used in a state-machine simulation that generates complete touches. The user study shows professional fencers split their preferences evenly between self-play touches and ground-truth touches, which the authors take as validation that the extracted strategy—including right-of-way dynamics—is correct.

Load-bearing premise

The load-bearing premise is that right-of-way can be inferred from which fencer's global displacement is larger by a fixed threshold of 0.3 meters, with no validation against actual referee calls; if this heuristic mislabels priority states, the transition tables and the entire strategy model learn from incorrect context.

Editorial extensions

If this is right

  • If the strategy model is correct, the same skill-cluster vocabulary and transition model can generate novel touches for any pairing of fencers, as demonstrated by self-play, play against a recorded real fencer, and interactive play against a professional.
  • The system provides a working interactive opponent for training purposes, with experts rating its understanding of priority, action diversity, and distance handling at 4, 4.25, and 3.75 out of 5 respectively.
  • The next-action prediction experiment shows the model's single-step choices are ranked between ground truth and random, while the full self-play chains are rated as good as ground truth, suggesting that the tactical quality emerges from the sequential composition rather than from any single action.
  • The termination conditions and priority-update rules allow the simulation to produce complete touches that respect the basic structure of real saber fencing, including scoring lights and right-of-way forfeiture on misses.

Reading between the lines

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

  • The single-step Markov assumption—conditioning only on the previous actions of both fencers—is likely the main bottleneck; the experts' comments about the system lacking the 'why' behind reactions suggest that extending the model to longer contexts or opponent-modeling would be a natural and testable improvement.
  • The priority heuristic (displacement difference with a fixed threshold of $\delta = 0.3$) is the most fragile link in the chain; replacing it with referee-verified right-of-way labels, which are knowable from official scoring, would directly test how much of the learned strategy depends on this assumption.
  • The success of the self-play evaluation suggests that broadcast video is a sufficient data source for tactical sports modeling; a similar pipeline could be applied to other rule-structured dueling sports with global positioning cues, though that remains an open extension rather than a claim of this paper.
  • The fixed 20-frame subclip windows may mask compound actions such as reprise or remise; an adaptive segmentation that lets one fencer take multiple actions while the opponent takes one could recover additional tactical richness and address the experts' noted limitation.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. VirtualFencer proposes a pipeline that reconstructs 3D fencing motion from in-the-wild broadcast videos, discovers low-level action clusters in an unsupervised manner, and models high-level fencing strategy as priority-conditioned transition probabilities over those clusters. The learned model is then used to generate new fencing touches via a state-machine simulation, and the paper evaluates the results through three user studies: next-action judgment, self-play bout ranking, and interactive play against professional fencers. The authors report that expert fencers cannot distinguish self-play touches from ground-truth touches, which they interpret as validation of the extracted strategy.

Significance. If the central claim holds, the paper would be a notable demonstration that a data-driven system can recover both low-level motion skills and high-level tactical structure from unlabeled video, without manual action annotation or motion capture. Strengths of the manuscript include a carefully designed domain-aware pipeline, a substantial video collection effort, and a user-study protocol with expert fencers that directly targets the realism of the generated behavior. The paper also is honest about several limitations, including cluster confusion due to body size and motion discontinuities in generated play. The main gaps are in the validation of the priority-annotation heuristic and in the statistical support for the user-study conclusions; these issues prevent the current version from fully establishing the strategy-extraction claim.

major comments (4)
  1. [Sec. 5.3, Sec. 6.3] The priority labels that condition the transition matrices in Eq. (1) are computed solely from the displacement threshold δ=0.3, with no agreement check against referee calls or expert ratings. In sabre, right-of-way can be held while retreating or lost after a miss, so a purely positional heuristic is likely to mislabel many 20-frame windows; each empirical distribution P_raw(u_t | u_{t-1}, v_{t-1}) is therefore estimated under an incorrect conditioning context. This is load-bearing because the claim of recovering true strategy depends on these labels. Please provide a validation of the priority labels (e.g., agreement statistics against a human expert or a referee) or an ablation with an alternative priority estimator.
  2. [Sec. 7.1, Table 2] Clustering accuracy is estimated from only 10 manually checked samples per cluster, and several clusters are barely above chance (C.10 at 0.4, C.21 at 0.5). The low-level skills are the vocabulary over which the strategy model is defined, so noisy clusters propagate into the transition statistics. Please report per-cluster accuracy with larger sample sizes and explain how poorly performing clusters are handled in the strategy model; this would materially strengthen the claim that the system recovers meaningful fencing actions.
  3. [Sec. 7.3, Table 1] The user study is presented as validation that experts 'split their preferences evenly' between VirtualFencer and ground truth, but the paper reports only raw vote counts and no significance test. A chi-square goodness-of-fit or an appropriate paired-comparison test is needed to support the claim. Without it, the observed counts (e.g., 218 first-place votes for VirtualFencer vs. 177 for ground truth in the second study) cannot be interpreted as 'even'; this is load-bearing because the paper's main validation message rests on this comparison.
  4. [Sec. 6.4.1] The simulation's priority update uses scoring-light metadata and manually selected finishing-action clusters in addition to the displacement rule, so the end-to-end user study does not isolate the Section 5.3 annotation. Consequently, the positive results in Sections 7.3.2 and 7.3.3 cannot validate the displacement-based priority heuristic, and the paper should either test the components separately or temper the claim that the strategy extraction is validated end-to-end.
minor comments (4)
  1. [Eq. (1)] Equation (1) uses a proportionality sign but does not explicitly define the normalization over actions; please state that sampling normalizes the product.
  2. [Sec. 5.1] The paper reports about 1.5 hours of fencing actions but does not state the number of touches or the number of 20-frame samples; including these statistics would make the dataset scale and the empirical counts more transparent.
  3. [Table 2] The term 'Off the line' in the action labels is not standard fencing terminology in this context; please clarify what this category denotes.
  4. [Sec. 7.3.3] The third user study involves only four participants; while this is a common exploratory setup, the qualitative conclusions should be presented as anecdotal rather than as statistically representative.

Circularity Check

1 steps flagged · score 5.0 of 10

Self-play 'validation' resamples the empirical transition distribution fitted to the same corpus; headline strategy-extraction evidence is in-sample, though the interactive expert study adds independent support.

  1. fitted input called prediction [Section 6.3 (Eq. 1), Section 6.4, Section 7.3.2, Abstract]
    "We use the fencing trajectories in our dataset to estimate the raw transition distribution. Given sequences of the form {(u0,v0,d0),(u1,v1,d1),...,(uT,vT,dT)}, we compute the raw transition distribution P_raw(u_t | u_{t-1}, v_{t-1}) from empirical frequency counts... users split their preferences evenly between touches produced by VirtualFencer’s self-play and ground truth touches, while clearly preferring both methods to the random baseline."

    The 'ground truth' touches and the next-action contexts used in the user studies come from the same 40-bout corpus (Sec. 5.1) from which P_raw is estimated by empirical frequency counts (Sec. 6.3). The simulator (Sec. 6.4) samples actions directly from that P_raw and retrieves clips from the same clusters. Consequently, a self-play touch is a Monte Carlo draw from the very distribution fitted to the training touches, so its first-order transition statistics match ground truth by construction. The headline result that experts cannot distinguish the two (Sec. 7.3.2) and the abstract's conclusion that the strategy extraction is 'validated' are therefore in-sample resampling checks, not independent predictions.

full rationale

The central derivation is a standard empirical transition model: P_raw is estimated by counting transitions in the collected videos, and simulation samples from it. That is not circular by itself. The circularity enters when the paper presents the self-play user study as validating the extracted strategy: because generated touches are sampled from P_raw and the 'ground truth' touches are from the same corpus used to estimate P_raw, the expert preference is an in-sample resampling fidelity check, statistically forced at the level of transition frequencies. The first 'next action prediction' study is similarly in-sample: the model is queried on contexts from the same dataset that supplied the empirical counts. The interactive professional-fencer study is a genuinely external interaction and gives the central claim some independent content, so this is partial rather than total circularity. The displacement-based priority heuristic (Sec. 5.3) is an additional unvalidated assumption that conditions both training and simulation, but it is a correctness risk rather than a by-construction reduction. No self-citation chain or imported uniqueness theorem is load-bearing.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The core contribution is a pipeline that fits empirical transition matrices and clusters to collected video data; the free parameters are hand-set thresholds and the modeling assumptions are stated but not independently validated. No new physical or theoretical entities are introduced.

free parameters (6)
  • delta (priority displacement threshold) = 0.3
    Hand-set threshold for classifying which fencer has initiative based on displacement difference; used to annotate priority mode in Section 5.3.
  • sigma (distance weighting sensitivity) = 0.5
    Tunable parameter in Eq. (1) that controls how strongly the Gaussian distance term weights transition probabilities; set by hand.
  • tau (crash distance threshold) = 1.5 meters
    Hand-set distance below which the simulation terminates as a collision (Section 6.4.2).
  • scoring distance threshold = 2 meters
    Hand-set distance for touch registered and terminal action termination conditions (Section 6.4.2).
  • action window length = 20 frames
    Based on the authors' measurement of typical action duration; used to segment clips into subclips for clustering (Section 6.1).
  • number of clusters = 30
    Chosen after experimentation; the authors report that 30 clusters balance granularity and redundancy (Section 7.1).
assumptions (5)
  • domain assumption The panning camera has fixed position and zoom, and the piste lines can be tracked to compute a homography that projects fencers into global coordinates.
    Section 5.2 relies on this to correct WHAM's global translations; if camera motion or zoom changes, the on-strip positions are wrong.
  • domain assumption Fencing actions are approximately 20 frames long, so 20-frame subclips capture single actions.
    Section 6.1 segments clips into 20-frame windows based on this measurement; if actions vary widely in duration, subclips mix multiple actions or split one action.
  • ad hoc to paper A single step of history is sufficient to model fencing transitions (Markov assumption).
    Section 6.3 hypothesizes 'a single step of history is sufficient'; this is a modeling assumption not validated against longer histories.
  • domain assumption The manually labeled cluster labels (e.g., advance, retreat, lunge) are reliable ground truth for evaluating clustering accuracy.
    Section 7.1 reports 85.67% clustering accuracy based on manual inspection of 10 samples per cluster, but the labels and the test samples are chosen by the same authors.
  • ad hoc to paper The displacement-based priority heuristic approximates real right-of-way.
    Section 5.3 defines priority from displacement differences with threshold delta=0.3; no comparison to referee decisions is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of VirtualFencer: Generating Fencing Bouts based on Strategies Extracted from In-the-Wild Videos." pith.science (2026). https://pith.science/paper/5VRU2M6U

@misc{pith2026250700261,
  author       = {Pith},
  title        = {Pith review of: VirtualFencer: Generating Fencing Bouts based on Strategies Extracted from In-the-Wild Videos},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5VRU2M6U}},
  note         = {Machine review of arXiv:2507.00261}
}
read the original abstract

Fencing is a sport where athletes engage in diverse yet strategically logical motions. While most motions fall into a few high-level actions (e.g. step, lunge, parry), the execution can vary widely-fast vs. slow, large vs. small, offensive vs. defensive. Moreover, a fencer's actions are informed by a strategy that often comes in response to the opponent's behavior. This combination of motion diversity with underlying two-player strategy motivates the application of data-driven modeling to fencing. We present VirtualFencer, a system capable of extracting 3D fencing motion and strategy from in-the-wild video without supervision, and then using that extracted knowledge to generate realistic fencing behavior. We demonstrate the versatile capabilities of our system by having it (i) fence against itself (self-play), (ii) fence against a real fencer's motion from online video, and (iii) fence interactively against a professional fencer.

Figures

Figures reproduced from arXiv: 2507.00261 by the authors.

Figure 1
Figure 1. Our system VirtualFencer leverages low-level skills discovered unsupervisedly from in-the-wild videos and fences [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overall architecture of VirtualFencer. We start with a large, unstructured dataset of in-the-wild fencing videos. Our [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Selected rows of the transition matrices, from top to bottom: [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: First User Study. Given the context of the in-progress touch, users are asked to rank the next action choices from [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Second User Study. Users were shown temporal motion rollouts for three conditions – from left to right: sampling [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Visualization of 25 overlay touches used in our user study of fencing against itself (self-play). Both left and right hand [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 33 canonical work pages

  1. [1]

    Tadeusz Bober, Alicja Rutkowska-Kucharska, Sebastian Jaroszczuk, Maciej Barabasz, and Wojciech Woznica. 2016. Original research papers. Kinematic characterisation of the lunge and the fleche in epee fencing: two case studies. Polish Journal of Sport and Tourism 23, 4 (2016), 181

    work page 2016

  2. [2]

    Zhe Cao, Tomas Simon, Shih-En Wei, and Yaser Sheikh. 2017. Realtime Multi- person 2D Pose Estimation Using Part Affinity Fields. In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . 1302–1310. doi:10.1109/CVPR. 2017.143

    doi:10.1109/cvpr 2017

  3. [3]

    Hongsuk Choi, Gyeongsik Moon, Ju Yong Chang, and Kyoung Mu Lee. 2021. Beyond Static Features for Temporally Consistent 3D Human Pose and Shape from a Video. In Conference on Computer Vision and Pattern Recognition (CVPR)

    work page 2021

  4. [4]

    Rohit Girdhar, Georgia Gkioxari, Lorenzo Torresani, Manohar Paluri, and Du Tran. 2018. Detect-and-track: Efficient pose estimation in videos. In Proceedings of the IEEE conference on computer vision and pattern recognition . 350–359

    work page 2018

  5. [5]

    Shubham Goel, Georgios Pavlakos, Jathushan Rajasegaran, Angjoo Kanazawa*, and Jitendra Malik*. 2023. Humans in 4D: Reconstructing and Tracking Humans with Transformers. In International Conference on Computer Vision (ICCV)

    work page 2023

  6. [6]

    Rıza Alp Güler, Natalia Neverova, and Iasonas Kokkinos. 2018. Densepose: Dense human pose estimation in the wild. In Proceedings of the IEEE conference on computer vision and pattern recognition . 7297–7306

    work page 2018

  7. [7]

    Kaiming He, Georgia Gkioxari, Piotr Dollár, and Ross Girshick. 2017. Mask r-cnn. In Proceedings of the IEEE international conference on computer vision . 2961–2969

    2017

  8. [8]

    Eldar Insafutdinov, Mykhaylo Andriluka, Leonid Pishchulin, Siyu Tang, Evgeny Levinkov, Bjoern Andres, and Bernt Schiele. 2017. Arttrack: Articulated multi- person tracking in the wild. In Proceedings of the IEEE conference on computer vision and pattern recognition . 6457–6465

    work page 2017

Show all 39 references
  1. [9]

    Eldar Insafutdinov, Leonid Pishchulin, Bjoern Andres, Mykhaylo Andriluka, and Bernt Schiele. 2016. DeeperCut: A Deeper, Stronger, and Faster Multi-person Pose Estimation Model. Springer International Publishing, 34–50. doi:10.1007/978-3- 319-46466-4_3

    2016 doi

  2. [10]

    Umar Iqbal, Anton Milan, and Juergen Gall. 2017. Posetrack: Joint multi-person pose estimation and tracking. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition . 2011–2020

    2017

  3. [11]

    Black, David W

    Angjoo Kanazawa, Michael J. Black, David W. Jacobs, and Jitendra Malik. 2018. End-to-end Recovery of Human Shape and Pose. In Computer Vision and Pattern Recognition (CVPR)

    2018

  4. [12]

    Zhang, Panna Felsen, and Jitendra Malik

    Angjoo Kanazawa, Jason Y. Zhang, Panna Felsen, and Jitendra Malik. 2019. Learning 3D Human Dynamics from Video. In Computer Vision and Pattern Recognition (CVPR)

    2019

  5. [13]

    Muhammed Kocabas, Nikos Athanasiou, and Michael J. Black. 2020. VIBE: Video Inference for Human Body Pose and Shape Estimation. In The IEEE Conference Zhiyin Lin, Purvi Goel, Joy Yun, C. Karen Liu, and Joao Pedro Araujo on Computer Vision and Pattern Recognition (CVPR)

    2020

  6. [14]

    Muhammed Kocabas, Ye Yuan, Pavlo Molchanov, Yunrong Guo, Michael Black, Otmar Hilliges, Jan Kautz, and Umar Iqbal. 2024. PACE: Human and Camera Motion Estimation from in-the-wild Videos. In International Conference on 3D Vision (3DV). arXiv:2310.13768

    2024 arXiv

  7. [15]

    Lucas Kovar, Michael Gleicher, and Frédéric Pighin. 2002. Motion graphs. ACM Trans. Graph. 21, 3 (July 2002), 473–482. doi:10.1145/566654.566605

    2002

  8. [16]

    Jehee Lee, Jinxiang Chai, Paul SA Reitsma, Jessica K Hodgins, and Nancy S Pollard. 2002. Interactive control of avatars animated with human motion data. In Proceedings of the 29th annual conference on Computer graphics and interactive techniques. 491–500

    2002

  9. [17]

    Jiefeng Li, Siyuan Bian, Qi Liu, Jiasheng Tang, Fan Wang, and Cewu Lu. 2023. NIKI: Neural Inverse Kinematics with Invertible Neural Networks for 3D Hu- man Pose and Shape Estimation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    2023

  10. [18]

    Jiefeng Li, Siyuan Bian, Chao Xu, Zhicun Chen, Lixin Yang, and Cewu Lu. 2023. HybrIK-X: Hybrid Analytical-Neural Inverse Kinematics for Whole-body Mesh Recovery. arXiv preprint arXiv:2304.05690 (2023)

    2023 arXiv

  11. [19]

    Alireza Golestaneh, and Kris M

    Zhengyi Luo, S. Alireza Golestaneh, and Kris M. Kitani. 2020. 3D Human Motion Estimation via Motion Compression and Refinement. In Proceedings of the Asian Conference on Computer Vision (ACCV)

    2020

  12. [20]

    Troje, Gerard Pons-Moll, and Michael J

    Naureen Mahmood, Nima Ghorbani, Nikolaus F. Troje, Gerard Pons-Moll, and Michael J. Black. 2019. AMASS: Archive of Motion Capture as Surface Shapes. In International Conference on Computer Vision . 5442–5451

    2019

  13. [21]

    Filip Malawski. 2018. Real-time first person perspective tracking and feedback system for weapon practice support in fencing. In Applications of Intelligent Systems. IOS Press, 79–88

    2018

  14. [22]

    Filip Malawski. 2022. Immersive feedback in fencing weapon practice using mixed reality. Computer Science 23, 1 (March 2022). doi:10.7494/csci.2022.23.1.4570

    2022 doi

  15. [23]

    Filip Malawski and Bogdan Kwolek. 2018. Recognition of action dynamics in fencing using multimodal cues. Image and Vision Computing 75 (2018), 1–10. doi:10.1016/j.imavis.2018.04.005

    2018 doi

  16. [24]

    Filip Malawski and Bogdan Kwolek. 2019. Improving multimodal action repre- sentation with joint motion history context. Journal of Visual Communication and Image Representation 61 (2019), 198–208. doi:10.1016/j.jvcir.2019.03.026

    2019 doi

  17. [25]

    Kevin C Moore, Frances ME Chow, and John YH Chow. 2015. Novel lunge biomechanics in modern Sabre fencing. Procedia engineering 112 (2015), 473– 478

    2015

  18. [26]

    Georgios Pavlakos, Vasileios Choutas, Nima Ghorbani, Timo Bolkart, Ahmed A. A. Osman, Dimitrios Tzionas, and Michael J. Black. 2019. Expressive Body Capture: 3D Hands, Face, and Body from a Single Image. In Proceedings IEEE Conf. on Computer Vision and Pattern Recognition (CVP...

    2019

  19. [27]

    Black, and Gül Varol

    Mathis Petrovich, Michael J. Black, and Gül Varol. 2023. TMR: Text-to-Motion Retrieval Using Contrastive 3D Human Motion Synthesis. In International Con- ference on Computer Vision (ICCV)

    2023

  20. [28]

    Henning, and Stefan Leutenegger

    Simon Schaefer, Dorian F. Henning, and Stefan Leutenegger. 2023. GloPro: Globally-Consistent Uncertainty-Aware 3D Human Pose Estimation & Tracking in the Wild. In 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). 3803–3810. doi:10.1109/IROS55552....

    2023

  21. [29]

    Soyong Shin, Juyong Kim, Eni Halilaj, and Michael J. Black. 2024. WHAM: Reconstructing World-grounded Humans with Accurate 3D Motion. InIEEE/CVF Conf. on Computer Vision and Pattern Recognition (CVPR)

    2024

  22. [30]

    Tomas Simon, Hanbyul Joo, Iain Matthews, and Yaser Sheikh. 2017. Hand Key- point Detection in Single Images Using Multiview Bootstrapping . In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . IEEE Computer Society, Los Alamitos, CA, USA, 4645–4653. doi:...

    2017 doi

  23. [31]

    Jie Song, Limin Wang, Luc Van Gool, and Otmar Hilliges. 2017. Thin-slicing network: A deep structured model for pose estimation in videos. In Proceedings of the IEEE conference on computer vision and pattern recognition . 4220–4229

    2017

  24. [32]

    Jonathan Taylor, Jamie Shotton, Toby Sharp, and Andrew Fitzgibbon. 2012. The vitruvian manifold: Inferring dense correspondences for one-shot human pose estimation. In 2012 IEEE conference on computer vision and pattern recognition . IEEE, 103–110

    2012

  25. [33]

    Wen-Li Wei, Jen-Chun Lin, Tyng-Luh Liu, and Hong-Yuan Mark Liao. 2022. Capturing Humans in Motion: Temporal-Attentive 3D Human Pose and Shape Estimation from Monocular Video. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

    2022

  26. [34]

    Vickie Ye, Georgios Pavlakos, Jitendra Malik, and Angjoo Kanazawa. 2023. De- coupling Human and Camera Motion from Videos in the Wild. InIEEE Conference on Computer Vision and Pattern Recognition (CVPR)

    2023

  27. [35]

    Haotian Zhang, Cristobal Sciutto, Maneesh Agrawala, and Kayvon Fatahalian

  28. [36]

    Hongwen Zhang, Yating Tian, Yuxiang Zhang, Mengcheng Li, Liang An, Zhenan Sun, and Yebin Liu. 2023. PyMAF-X: Towards Well-aligned Full-body Model Regression from Monocular Images. IEEE Transactions on Pattern Analysis and Machine Intelligence (2023)

    2023

  29. [37]

    Haotian Zhang, Ye Yuan, Viktor Makoviychuk, Yunrong Guo, Sanja Fidler, Xue Bin Peng, and Kayvon Fatahalian. [n. d.]. Learning Physically Simulated Tennis Skills from Broadcast Videos. ACM Trans. Graph. ([n. d.]), 14 pages. doi:10.1145/3592408

    doi

  30. [38]

    It understands right of way, and it knows when to attack and defend. This logic takes most beginning fencers many months to properly learn

    Kevin Zhu, Alexander Wong, and John McPhee. 2022. FenceNet: Fine-Grained Footwork Recognition in Fencing. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops . 3589–3598. VirtualFencer: Generating Fencing Bouts based on Strateg...

    2022

  31. [2021]

    ACM Trans

    Vid2Player: Controllable Video Sprites That Behave and Appear Like Professional Tennis Players. ACM Trans. Graph. 40, 3, Article 24 (May 2021), 16 pages. doi:10.1145/3448978

    2021 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.