Pith. sign in

REVIEW 3 major objections 50 references

The first large multimodal 4D human–dog dataset enables models that generate realistic pet motion from human gestures and voice.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-14 12:53 UTC pith:QUGC3UER

load-bearing objection First real multimodal 4D human–dog interaction dataset at multi-million-frame scale; generation model is competent but secondary, and dog reconstruction quality remains unquantified. the 3 major comments →

arxiv 2607.10287 v1 pith:QUGC3UER submitted 2026-07-11 cs.CV

InterPet4D: A Multimodal 4D Human-Pet Interaction Dataset for Pet Motion Generation

Yichen Peng , Jyun-Ting Song , Chen-Chieh Liao , Kris Kitani , Hideki Koike , Erwin Wu This is my paper
classification cs.CV
keywords human-pet interaction4D datasetdog motion generationmultimodal captureMotionGPTPetVAESMALgesture-conditioned generation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Human–pet interaction has lagged behind human–human and human–object work because no large, synchronized 3D dataset existed. This paper fills that gap with InterPet4D: 6.8 million frames of 13 dogs of 11 breeds interacting with 23 people under a controlled multi-view and egocentric capture setup, with full annotations for video, audio, 2D/3D keypoints, and parametric meshes. On top of the data the authors introduce InterPetMoGen, a MotionGPT-style model that tokenizes human body, hand, and audio signals and autoregressively generates dog motion. The model reaches an FID of 11.21 and clearly beats Seq2Seq and diffusion baselines, showing that the dataset is large and rich enough to learn realistic cross-species responses.

Core claim

InterPet4D is the first large-scale multimodal 4D dataset of natural human–dog interactions (6.8 M frames, 13 dogs, 23 humans), and the InterPetMoGen model trained on it can generate plausible 3D dog motion conditioned on human body/hand gestures and audio, achieving FID 11.21 and outperforming Seq2Seq and DiT baselines.

What carries the argument

InterPetMoGen (IPMG): a PetVAE tokenizer that discretizes dog pose (with bone-length consistency) plus a modality-aware attention (MAA) prefix-LM that conditions autoregressive pet-motion tokens on bidirectional human body/hand tokens and cross-attended audio.

Load-bearing premise

The multi-view keypoint triangulation and SMAL fitting pipeline, which freezes dog shape from one frame and masks out human-occluded keypoints, must produce accurate enough 3D dog motion that the learned responses are real rather than reconstruction artifacts.

What would settle it

A controlled hold-out of interaction sequences whose 3D dog reconstructions are independently verified by multi-view marker or high-precision optical systems, then checking whether InterPetMoGen’s FID and user-study scores remain superior to the same baselines when trained only on the verified subset.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 0 minor

Summary. The paper introduces InterPet4D, claimed as the first large-scale multimodal 4D dataset of human–dog interactions (6.8M frames, 12 multi-view + egocentric RGB, audio, 2D/3D keypoints, SMPL-X/MANO and SMAL meshes) collected from 23 humans and 13 dogs of 11 breeds under a four-category protocol (petting, commanding, calling, free-form). It further proposes InterPetMoGen (IPMG): separate VQ-VAE tokenizers for body, hand, pet pose (PetVAE with bone-length auxiliary loss), and root translation, plus MERT audio features, fed to a prefix-LM GPT with modality-aware attention (bidirectional human conditioning, causal coarse-to-fine translation/pet decoding). On a fixed 80/20 per-dog split, IPMG reports FID_k/s of 11.21/12.96 and higher R-Precision/diversity than Seq2Seq-Transformer and DiT baselines (Table 3), with supporting modality/PetVAE ablations (Tables 4–5, Supp. E), qualitative sequences (Fig. 4), and a 12-participant Likert study with ANOVA (Table 6).

Significance. If the reconstructions are sufficiently accurate, InterPet4D fills a genuine gap: existing interaction datasets are human–human or human–object, and animal datasets lack synchronized human motion and audio. A public multimodal 4D resource with a clear taxonomy and a reproducible generation baseline would enable cross-species interaction modeling for animation, robotics, and behavioral analysis. Strengths include the multi-sensor capture design (lower cameras for pets, egocentric glasses), the hierarchical MAA design that matches the interaction structure, consistent gains across FID, R-Precision, diversity, and a small but statistically tested user study, and the planned Hugging Face release. The work is therefore of clear community value provided reconstruction fidelity is demonstrated.

major comments (3)
  1. Section 4.2 (and the central claim that FID 11.21 demonstrates realistic interaction modeling): dog 3D targets are obtained by multi-view HRNet triangulation after Mask-R-CNN human-mask filtering, RTS smoothing, and SMAL fitting with shape parameters frozen from a single manually chosen canonical frame. No MPJPE, PCK, multi-view reprojection error, or contact-consistency numbers are reported, despite the paper’s own emphasis on severe close-range occlusion. Because the same sequences train PetVAE/GPT and the dog-identity classifier used for FID features (Supp. F), systematic reconstruction bias is shared by targets, model, and metric. Quantitative reconstruction validation (at least multi-view reprojection error stratified by interaction category, and preferably a small marker-based or multi-view consistency check) is required before the generation numbers can be attributed to interactio
  2. Table 3 / Supp. F: the FID feature extractor is a 12-class dog-identity ResNet trained on the same reconstructed motion data. No external validation (e.g., correlation with MPJPE, velocity error, or human preference) is provided, so the large reported FID reductions (47% vs Seq2Seq, 83% vs DiT) remain only partially interpretable. Either validate the feature space against independent motion quality measures or report additional metrics (e.g., bone-length error, foot-skate, command-response latency) that do not rely on this classifier.
  3. Section 6.1 and Table 1: total training motion after 30 fps downsampling and 10 s windowing is only ~87.5 minutes, with 200 train / 40 val clips and all dogs appearing in both splits. While the per-dog split avoids complete identity leakage, the absolute scale is small for a 28–30 M parameter autoregressive model and for claims of “large-scale” diversity across 11 breeds. The manuscript should discuss overfitting risk more carefully and, if possible, report leave-one-dog-out or leave-one-breed-out numbers to show that the learned p(P|H,A) generalizes beyond the reconstruction idiosyncrasies of the training animals.

Circularity Check

1 steps flagged

Empirical dataset+model paper with only mild metric self-reference; no derivation reduces to its inputs by construction.

specific steps
  1. other [Supp. F (Dog Motion FID Model) + Sec. 6.2/Table 3]
    "Following the standard practice of using a pretrained classifier as a feature extractor for Fréchet Inception Distance (FID), we train a dog identity classifier on our motion data to serve as the backbone for motion FID computation. ... After training, we discard the classification head and use the penultimate 512-dimensional features to compute FID between the generated and ground-truth motion distributions"

    The FID backbone is trained on the identical reconstructed pet-motion sequences that serve as both training targets for PetVAE/GPT and the ground-truth distribution for the metric. Systematic reconstruction biases are therefore shared by model and evaluator. This is only mild self-reference (not a fitted parameter renamed as prediction, and relative rankings remain informative), but it is the sole circular-adjacent step.

full rationale

InterPet4D/InterPetMoGen is a standard dataset-plus-generative-model paper. The central claims (new multimodal capture of human–dog interactions; IPMG with PetVAE+MAA outperforms Seq2Seq and DiT on FID/R-Precision/Diversity and a user study) are empirical comparisons on held-out clips, not first-principles derivations. No equation defines a quantity in terms of the quantity it is said to predict; no parameter is fitted to a subset and then reported as an independent prediction of a closely related quantity; no uniqueness theorem or ansatz is imported via self-citation to force the architecture. The sole mild self-reference is that the FID feature extractor (Supp. F) is a dog-identity classifier trained on the same reconstructed motion sequences used as generation targets; this is common practice for motion FID and does not make the reported ranking of models tautological, because all baselines are evaluated under the identical fixed metric and the user-study ratings supply an independent perceptual check. Reconstruction quality itself is unquantified, but that is a validity concern, not circularity. Score 1 reflects only the minor shared-data FID backbone; the derivation chain is otherwise self-contained.

Axiom & Free-Parameter Ledger

6 free parameters · 5 axioms · 3 invented entities

The work is empirical: it rests on standard parametric body models, off-the-shelf detectors, and a conventional VQ-VAE + GPT pipeline. Free parameters are the usual training hyper-parameters and the frozen SMAL shape vectors. No new physical entities are postulated; the main domain assumptions are that the chosen interaction protocol and reconstruction pipeline adequately sample natural human–dog dynamics.

free parameters (6)
  • VQ codebook size = 512
    Shared codebook size of 512 for all four VQ-VAEs; chosen by authors, not derived.
  • Temporal downsampling factor = 4×
    4× temporal downsampling in PetVAE / human / translation tokenizers; design choice that sets token rate.
  • GPT model size / depth = 6 blocks, ~30 M
    28–30.3 M parameter prefix-LM with 6 blocks, 8 heads; architecture hyper-parameter.
  • SMAL shape parameters (per dog) = per-dog manual fit
    Shape β frozen from a single manually fitted canonical frame per dog (Section 4.2); not optimized per frame.
  • Commitment loss weight β = 0.02
    β = 0.02 with linear warm-up over 50 epochs for VQ training.
  • Clip length / sliding-window stride = 300 / 50
    T = 300 frames (10 s) with 50-frame stride; determines training distribution.
axioms (5)
  • domain assumption SMPL / SMPL-X / MANO and SMAL parametric models are adequate low-dimensional representations of human and dog shape/pose for the captured interactions.
    Invoked throughout Section 4; reconstruction quality is never validated against marker-based ground truth.
  • domain assumption Multi-view triangulation of 2D keypoints (HRNet / WiLoR) after human-mask filtering yields sufficiently accurate 3D joints under heavy mutual occlusion.
    Core of Sections 4.1–4.2; no quantitative reconstruction error reported.
  • domain assumption The four interaction categories (petting, commanding, calling, free-form) plus trained-dog obedience cover the distribution needed for realistic human–pet motion generation.
    Section 3.2 protocol design; free-form sessions mitigate but do not eliminate selection bias toward trained behaviors.
  • domain assumption Discrete VQ tokens plus autoregressive next-token prediction can capture the one-to-many conditional distribution p(P|H,A).
    Standard MotionGPT modeling choice adopted in Section 5.
  • standard math Standard VQ-VAE and transformer training mathematics (commitment loss, cross-entropy, etc.).
    Used without modification in Sections 5.3–5.4.
invented entities (3)
  • InterPet4D dataset no independent evidence
    purpose: Provide the first large-scale multimodal 4D human–dog interaction corpus.
    New data resource; independent evidence will exist once the Hugging Face release is public and usable by others.
  • PetVAE tokenizer no independent evidence
    purpose: Learn a discrete latent space for dog pose that preserves bone lengths.
    Standard VQ-VAE specialized to dog skeleton; no claim of a new physical quantity.
  • Modality-Aware Attention (MAA) mask no independent evidence
    purpose: Enforce coarse-to-fine generation hierarchy (human bidirectional, translation/pet causal).
    Architectural mask; ablated in Table 3 but not an independent scientific entity.

pith-pipeline@v1.1.0-grok45 · 21499 in / 3691 out tokens · 29457 ms · 2026-07-14T12:53:39.978941+00:00 · methodology

0 comments
read the original abstract

Human-pet interaction estimation and generation remain underexplored due to the absence of a high-quality large-scale dataset. We present InterPet4D, the first multimodal dataset capturing natural interactions between humans and dogs. Using a synchronized multi-view capture system, we record human-dog obedience tasks and provide annotations for both humans and dogs, including multi-view and egocentric videos, segmentations, 2D and 3D keypoints, meshes, and audio tracks. InterPet4D consists of 6.8 million frames collected from 13 dogs of 11 breeds interacting with 23 human participants. We further introduce the InterPetMoGen framework for human-pet interaction motion generation. Our proposed model achieves an FID score of 11.21 and substantially outperforms the Seq2Seq and DiT baselines, demonstrating the effectiveness of InterPet4D for modeling realistic human-pet interactions.

Figures

Figures reproduced from arXiv: 2607.10287 by Chen-Chieh Liao, Erwin Wu, Hideki Koike, Jyun-Ting Song, Kris Kitani, Yichen Peng.

Figure 1
Figure 1. Figure 1: The InterPet4D multimodal Human-Pet Interaction Dataset. Abstract. Human-pet interaction estimation and generation remain un￾derexplored due to the absence of high-quality large-scale dataset. We present InterPet4D, the first multimodal dataset capturing natural in￾teractions between humans and dogs. Using a synchronized multi-view capture system, we record human–dog obedience tasks and provide an￾notation… view at source ↗
Figure 2
Figure 2. Figure 2: Data collection setup [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: InterPetMoGen architecture. Human gesture tokens and audio features are encoded by modality-specific encoders and fused through an autoregressive trans￾former. The model generates global translation and pet motion tokens conditioned on human body, hand, and audio tokens. The generated tokens are decoded into continu￾ous dog motion using the PetVAE decoder. 5.2 Model Overview As illustrated in [PITH_FULL_I… view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative results. We visualize InterPetMoGen outputs in a sequential command of "come->jump->turn" compared to other baselines. For each example, only the trainer’s hands are rendered for better visibility of the dog. 6.6 Qualitative Results [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Dog skeleton topology. Our 20-keypoint skeleton covers the facial region, torso, limbs, and tail base. Autoregressive Transformer (GPT). The prefix-LM GPT model has 30.3M pa￾rameters and consists of 6 transformer blocks. Each block contains a self-attention layer with 8 heads, a cross-attention layer for audio conditioning, and a feed￾forward network with hidden dimension 2048. The input sequence consists … view at source ↗
Figure 6
Figure 6. Figure 6: Shape and motion analysis. (a) Mean bone lengths differ markedly across the 12 breeds, capturing diverse body proportions. (b) per clip bone length vectors form tight per dog clusters in t-SNE space, demonstrating consistent and accurate shape estimation. (c) Motion features are broadly distributed and overlapping across breeds, reflecting the rich motion diversity in our dataset. Dog and Labrador Retrieve… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

50 extracted references · 2 canonical work pages

  1. [1]

    Agrawal, V., et al.: Seamless interaction: Dyadic audiovisual motion modeling and large-scale dataset (2025),https://arxiv.org/abs/2506.22554

    Pith/arXiv arXiv 2025

  2. [2]

    In: CVPR

    Bhatnagar, B.L., Xie, X., Petrov, I.A., Sminchisescu, C., Theobalt, C., Pons-Moll, G.: BEHAVE: Dataset and method for tracking human object interactions. In: CVPR. pp. 15935–15946 (2022)

    2022

  3. [3]

    In: ECCV

    Biggs, B., Bober, O., Sherrill, J.M., Koepke, A.S., Mayol-Cuevas, W.W., Sherrill, D.M., Sherrill, S.M., Sherrill, J.M.: Who left the dogs out? 3D animal reconstruc- tion with expectation maximization in the loop. In: ECCV. pp. 195–211 (2020)

    2020

  4. [4]

    In: Asian Conference on Computer Vision

    Biggs, B., Roddick, T., Fitzgibbon, A., Cipolla, R.: Creatures great and smal: Recovering the shape and motion of animals from video. In: Asian Conference on Computer Vision. pp. 3–19. Springer (2018)

    2018

  5. [5]

    Cai, Y., Wu, Y., Li, K., Zhou, Y., Zheng, B., Liu, H.: Flooddiffusion: Tailored diffusion forcing for streaming motion generation (2026),https://arxiv.org/ abs/2512.03520

    arXiv 2026

  6. [6]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Cao, J., Tang, H., Fang, H.S., Shen, X., Lu, C., Tai, Y.W.: Cross-domain adapta- tion for animal pose estimation. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 9498–9507 (2019)

    2019

  7. [7]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Chen, X., Jiang, B., Liu, W., Huang, Z., Fu, B., Chen, T., Yu, G.: Executing your commands via motion diffusion in latent space. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 18000–18010 (2023)

    2023

  8. [8]

    In: Proceed- ings of the IEEE/CVF conference on computer vision and pattern recognition

    Cheng, B., Xiao, B., Wang, J., Shi, H., Huang, T.S., Zhang, L.: Higherhrnet: Scale- aware representation learning for bottom-up human pose estimation. In: Proceed- ings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 5386–5395 (2020) 16 Y. Peng et al

    2020

  9. [9]

    In: CVPR

    Fan, Z., Taheri, O., Tzionas, D., Kocabas, M., Kaufmann, M., Black, M.J., Hilliges, O.: ARCTIC: A dataset for dexterous bimanual hand-object manipulation. In: CVPR. pp. 12943–12954 (2023)

    2023

  10. [10]

    Guo, C., Mu, Y., Javed, M.G., Wang, S., Cheng, L.: Momask: Generative masked modeling of 3d human motions (2023),https://arxiv.org/abs/2312.00063

    Pith/arXiv arXiv 2023

  11. [11]

    In: Proceedings of the IEEE international conference on computer vision

    He, K., Gkioxari, G., Dollár, P., Girshick, R.: Mask r-cnn. In: Proceedings of the IEEE international conference on computer vision. pp. 2961–2969 (2017)

    2017

  12. [12]

    Heusel,M.,Ramsauer,H.,Unterthiner,T.,Nessler,B.,Hochreiter,S.:Ganstrained by a two time-scale update rule converge to a local nash equilibrium (2018),https: //arxiv.org/abs/1706.08500

    Pith/arXiv arXiv 2018

  13. [13]

    Higami, A., Oshima, K., Shiramatsu, T.I., Takahashi, H., Nobuhara, S., Nishino, K.: Ratbodyformer: Rat body surface from keypoints (2025),https://arxiv.org/ abs/2412.09599

    Pith/arXiv arXiv 2025

  14. [14]

    Machines11(7), 677 (2023)

    Hussain, M.: Yolo-v1 to yolo-v8, the rise of yolo and its complementary nature toward digital manufacturing and industrial defect detection. Machines11(7), 677 (2023)

    2023

  15. [15]

    Advances in Neural Information Processing Systems36 (2024)

    Jiang, B., Chen, X., Liu, W., Yu, J., Yu, G., Chen, T.: Motiongpt: Human motion as a foreign language. Advances in Neural Information Processing Systems36 (2024)

    2024

  16. [16]

    In: European Conference on Computer Vision

    Jin, S., Xu, L., Xu, J., Wang, C., Liu, W., Qian, C., Ouyang, W., Luo, P.: Whole- body human pose estimation in the wild. In: European Conference on Computer Vision. pp. 196–214. Springer (2020)

    2020

  17. [17]

    Advances in Neural Information Process- ing Systems37, 107270–107285 (2024)

    Khirodkar,R.,Song,J.T.,Cao,J.,Luo,Z.,Kitani,K.:Harmony4d:Avideodataset for in-the-wild close human interactions. Advances in Neural Information Process- ing Systems37, 107270–107285 (2024)

    2024

  18. [18]

    Li, Y., Yuan, R., Zhang, G., Ma, Y., Chen, X., Yin, H., Lin, C., Ragni, A., Benetos, E., Gyenge, N., Dannenberg, R., Liu, R., Chen, W., Xia, G., Shi, Y., Huang, W., Guo, Y., Fu, J.: Mert: Acoustic music understanding model with large-scale self- supervised training (2023)

    2023

  19. [19]

    In: IJCV (2024)

    Liang, H., Zhang, W., Li, W., Yu, J., Xu, L.: InterGen: Diffusion-based multi- human motion generation under complex interactions. In: IJCV (2024)

    2024

  20. [20]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Liu, H., Zhu, Z., Becherini, G., Peng, Y., Su, M., Zhou, Y., Zhe, X., Iwamoto, N., Zheng, B., Black, M.J.: Emage: Towards unified holistic co-speech gesture generation via expressive masked audio gesture modeling. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 1144–1154 (June 2024)

    2024

  21. [21]

    In: Computer Vision – ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part VII

    Liu, H., Zhu, Z., Iwamoto, N., Peng, Y., Li, Z., Zhou, Y., Bozkurt, E., Zheng, B.: Beat: A large-scale semantic and emotional multi-modal dataset for conversational gestures synthesis. In: Computer Vision – ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part VII. p. 612–630. Springer- Verlag, Berlin, Heidelberg (2...

    work page doi:10.1007/978-3-031-20071- 2022

  22. [22]

    ACM Transactions on Graphics (TOG)44(4), 1–12 (2025)

    Liu, I., Xu, Z., Wang, Y., Tan, H., Xu, Z., Wang, X., Su, H., Shi, Z.: Riganything: Template-free autoregressive rigging for diverse 3d assets. ACM Transactions on Graphics (TOG)44(4), 1–12 (2025)

    2025

  23. [23]

    ACM TOG34(6), 1–16 (2015)

    Loper, M., Mahmood, N., Romero, J., Pons-Moll, G., Black, M.J.: SMPL: A skinned multi-person linear model. ACM TOG34(6), 1–16 (2015)

    2015

  24. [24]

    Lyu, C., Zhang, W., Huang, H., Zhou, Y., Wang, Y., Liu, Y., Zhang, S., Chen, K.: Rtmdet: An empirical study of designing real-time object detectors (2022) InterPet4D 17

    2022

  25. [25]

    Nature Neuroscience21(9), 1281–1289 (2018)

    Mathis, A., Mamidanna, P., Cull, K.M., Hainmueller, B., Hosp, S., Liao, C.L., Bethge, M.: DeepLabCut: Markerless pose estimation of user-defined body parts with deep learning. Nature Neuroscience21(9), 1281–1289 (2018)

    2018

  26. [26]

    In: CVPR

    Müller, L., Ye, V., Pavlakos, G., Black, M.J., Kanazawa, A.: BUDDI: Building dynamic human-human interaction. In: CVPR. pp. 978–988 (2024)

    2024

  27. [27]

    Ng, X.L., Ong, K.E., Zheng, Q., Ni, Y., Yeo, S.Y., Liu, J.: Animal kingdom: A large anddiversedatasetforanimalbehaviorunderstanding.In:CVPR.pp.19023–19034 (2022)

    2022

  28. [28]

    van den Oord, A., Vinyals, O., Kavukcuoglu, K.: Neural discrete representation learning (2018),https://arxiv.org/abs/1711.00937

    Pith/arXiv arXiv 2018

  29. [29]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Pavlakos, G., Choutas, V., Ghorbani, N., Bolkart, T., Osman, A.A., Tzionas, D., Black, M.J.: Expressive body capture: 3d hands, face, and body from a single im- age. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 10975–10985 (2019)

    2019

  30. [30]

    Peng, Y., Song, J.T., Jung, S., Liu, R., Liu, H., Chu, X., Liu, R., Wu, E., Koike, H., Kitani, K.: Dyadit: A multi-modal diffusion transformer for socially favorable dyadic gesture generation (2026),https://arxiv.org/abs/2602.23165

    arXiv 2026

  31. [31]

    In: ICCV

    Petrovich, M., Black, M.J., Varol, G.: Action-conditioned 3D human motion syn- thesis with transformer VAE. In: ICCV. pp. 10985–10995 (2021)

    2021

  32. [32]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Potamias, R.A., Zhang, J., Deng, J., Zafeiriou, S.: Wilor: End-to-end 3d hand localization and reconstruction in-the-wild. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 12242–12254 (2025)

    2025

  33. [33]

    arXiv preprint arXiv:2201.02610 (2022)

    Romero, J., Tzionas, D., Black, M.J.: Embodied hands: Modeling and capturing hands and bodies together. arXiv preprint arXiv:2201.02610 (2022)

    Pith/arXiv arXiv 2022

  34. [34]

    In: CVPR

    Rüegg, N., Zuffi, S., Schindler, K., Black, M.J.: BARC: Learning to regress 3D dog shape from images with reinforcement of breed-specific 3D shape priors. In: CVPR. pp. 3886–3896 (2022)

    2022

  35. [35]

    In: CVPR

    Rüegg, N., Zuffi, S., Schindler, K., Black, M.J.: BITE: Beyond priors for improved three-D dog pose estimation. In: CVPR. pp. 8867–8876 (2023)

    2023

  36. [36]

    In: Computer Vision – ECCV 2024: 18th Eu- ropean Conference, Milan, Italy, September 29–October 4, 2024, Proceedings, Part LXXIX

    Sabathier, R., Mitra, N.J., Novotny, D.: Animal avatars: Reconstructing animat- able 3d animals from casual videos. In: Computer Vision – ECCV 2024: 18th Eu- ropean Conference, Milan, Italy, September 29–October 4, 2024, Proceedings, Part LXXIX. p. 270–287 (2024).https://doi.org/10.1007/978-3-031-72986-7_16

    work page doi:10.1007/978-3-031-72986-7_16 2024

  37. [37]

    IEEE transactions on auto- matic control53(3), 845–849 (2008)

    Särkkä, S.: Unscented rauch–tung–striebel smoother. IEEE transactions on auto- matic control53(3), 845–849 (2008)

    2008

  38. [38]

    arXiv preprint arXiv:2601.21337 (2026)

    Shi, X., Wang, X., Guo, Z., Wang, Y., Zhang, P., Zhang, X., Guo, Z., Hao, H., Xi, Y., Yang, B., Xu, J., Zhou, J., Lin, J.: Qwen3-asr technical report. arXiv preprint arXiv:2601.21337 (2026)

    Pith/arXiv arXiv 2026

  39. [39]

    In: Thirteenth International Conference on 3D Vision

    Shin, S., Lee, C., Chen, H., Song, J.T., Halilaj, E., Kitani, K.: Bodycontact4d: A multi-view video dataset for understanding human and environment interactions. In: Thirteenth International Conference on 3D Vision

  40. [40]

    CVPR (2023)

    Sinha, S., Shapovalov, R., Reizenstein, J., Rocco, I., Neverova, N., Vedaldi, A., Novotny, D.: Common pets in 3d: Dynamic new-view synthesis of real-life de- formable categories. CVPR (2023)

    2023

  41. [41]

    Siyao, L., Gu, T., Yang, Z., Lin, Z., Liu, Z., Ding, H., Yang, L., Loy, C.C.: Duolando: Follower gpt with off-policy reinforcement learning for dance accom- paniment (2024),https://arxiv.org/abs/2403.18811

    Pith/arXiv arXiv 2024

  42. [42]

    In: Thirteenth International Conference on 3D Vision 18 Y

    Song, J.T., Kim, J., Cao, J., Lei, Y., Yagi, T., Kitani, K.: Contact4d: A video dataset for whole-body human motion and finger contact in dexterous operations. In: Thirteenth International Conference on 3D Vision 18 Y. Peng et al

  43. [43]

    In: ECCV

    Taheri, O., Ghorbani, N., Black, M.J., Tzionas, D.: GRAB: A dataset of whole- body human grasping of objects. In: ECCV. pp. 581–600 (2020)

    2020

  44. [44]

    In: ICLR (2023)

    Tevet, G., Raab, S., Gordon, B., Shafir, Y., Cohen-Or, D., Bermano, A.H.: Human motion diffusion model. In: ICLR (2023)

    2023

  45. [45]

    Wang, Z., Chen, S., Mo, L., Gao, X., Shen, Y., Ding, L., Liang, W.: Dogmo: A large-scale multi-view rgb-d dataset for 4d canine motion recovery (2025),https: //arxiv.org/abs/2510.24117

    arXiv 2025

  46. [46]

    In: 2017 IEEE international conference on image processing (ICIP)

    Wojke, N., Bewley, A., Paulus, D.: Simple online and realtime tracking with a deep association metric. In: 2017 IEEE international conference on image processing (ICIP). pp. 3645–3649. IEEE (2017)

    2017

  47. [47]

    In: CVPR

    Yi, H., Liang, H., Liu, Y., Cao, Q., Wen, Y., Bolkart, T., Tao, D., Black, M.J.: Generating holistic 3D human motion from speech. In: CVPR. pp. 469–480 (2023)

    2023

  48. [48]

    In: CVPR

    Zhang, J., Zhang, Y., Cun, X., Huang, S., Zhang, Y., Zhao, H., Lu, H., Shen, X.: T2M-GPT: Generating human motion from textual descriptions with discrete representations. In: CVPR. pp. 14141–14150 (2023)

    2023

  49. [49]

    IEEE TPAMI46(4), 2514–2527 (2024)

    Zhang, M., Cai, Z., Pan, L., Hong, F., Guo, X., Yang, L., Liu, Z.: MotionDiffuse: Text-driven human motion generation with diffusion model. IEEE TPAMI46(4), 2514–2527 (2024)

    2024

  50. [50]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Zuffi, S., Kanazawa, A., Jacobs, D.W., Black, M.J.: 3d menagerie: Modeling the 3d shape and pose of animals. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 6365–6373 (2017) InterPet4D 19 This supplementary material contains the following sections: – A.Network Architecture & Training Details – B.Dog Skeleton Definiti...

    2017