Pith. sign in

REVIEW 2 major objections 7 minor 1 cited by

RatBodyFormer: Rat Body Surface from Keypoints

T0 review · 2 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Ten sparse keypoints can be turned into a dense 3D rat body surface.

desk verdict First dense rat body-surface dataset plus a workable keypoint-to-surface transformer; the marker-based results are honest, but the abstract overclaims masked learning and D3's markerless numbers lean on silhouettes that also supervise the test-time fit. read the letter →

arxiv 2412.09599 v3 pith:O2WFY4DY submitted 2024-12-12 cs.CV

Ayaka Higami , Karin Oshima , Tomoyo Isoguchi Shiramatsu , Hirokazu Takahashi , Shohei Nobuhara , Ko Nishino This is my paper · ORCID
classification cs.CV
keywords ratbehavioranalysis3Dbodysurfacereconstructionkeypoint-to-surfaceregressionmultiviewcapturesystemtransformerencoder-decodernon-rigiddeformationanimalposeestimationsemi-automaticannotation
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

RatBodyFormer is a method for turning the ten 3D keypoints that standard detectors already find on a rat—nose, eyes, ears, paws, tail base—into a dense cloud of body-surface points. The paper argues that the rat's featureless, highly non-rigid fur coat, which keypoint detectors cannot latch onto, can be modeled as a learned deformation of a canonical reference surface. To learn that deformation it introduces the RatDome capture system and dataset, in which colored beads and paint temporarily provide trackable surface features during training only. On real captures the method achieves mean $L^2$ error around 6.5 mm, close to the bead diameter, and outperforms a prior whole-body mesh-fitting approach by a large margin. The authors present this as the first foundation for dense, marker-free body-surface reconstruction of rats for automated behavior analysis.

What carries the argument

The central object is a canonical 3D body surface: a reference standing-on-two-feet surface onto which all bead and paint marker positions are mapped with as-rigid-as-possible deformation. RatBodyFormer is a transformer encoder-decoder whose encoder tokens are the normalized displacement of each detected keypoint from its canonical position, and whose decoder queries are the normalized displacements of body-surface points from an analytically deformed initial guess; the decoder outputs corrected surface-point displacements. The surface points are therefore not indexed by fixed vertex identity, which lets one training set absorb differently-beaded rats. Per-rat scale and translation parameters are refined during training and at inference time using silhouette and position-consistency losses, and the model is trained with masked-learning plus semi-automatically annotated frames generated by the model itself.

What would settle it

Record two different behaviors, such as curling versus stretching, in which the ten 3D keypoints are at nearly identical positions but the marker-covered body surfaces differ by more than the reported mean error; if those frames are held out, the deterministic keypoint-to-surface mapping will mispredict one of the two surfaces and the error will exceed the roughly 6.5 mm average.

Watch

Extended reading notes

Core claim

The paper claims that dense 3D body-surface reconstruction of a freely moving rat is achievable from sparse detected keypoints alone, without any image or silhouette input at inference time. Its core result is that a transformer encoder-decoder, trained on paired keypoint and surface-point data collected from marker-attached rats, can regress body-surface point displacements relative to a canonical surface with mean $L^2$ error of roughly 6.5 mm, comparable to the diameter of the training markers, and generalizes to unseen rats aged 5 to 14 weeks. The same architecture also supports a keypoint-driven animatable rat model. The paper positions this as the first method of its kind for rats, with accuracy well above the preceding mouse-mesh-based baseline.

Load-bearing premise

The load-bearing premise is that the body surface is a single deterministic function of the ten keypoint positions, so any pose in which two genuinely different body-surface shapes share the same keypoint configuration is outside what the trained network can represent.

Editorial extensions

If this is right

  • A rat's dense body surface, not just its skeleton, becomes a measurable behavioral signal, so curling, stretching, and subtle torso changes can be quantified.
  • Because the surface is predicted from the same ten keypoints that existing detectors output, the method can be applied to keypoint-only recordings without any new markers.
  • The animatable surface model can synthesize novel views and poses, enabling analysis-by-synthesis and virtual-stimulus experiments.
  • The capture protocol generalizes across rat ages and body shapes, covering the 5-to-14-week age range common in biomedical experiments.
  • Semi-automatic annotation reduces the manual labeling burden for building paired keypoint-surface datasets for other featureless animals.

Reading between the lines

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

  • Editorial: Conditioning the same architecture on images or silhouettes in addition to keypoints is the natural next step; the paper itself notes that keypoints alone underdetermine the surface in complex interactions, so visual cues should resolve that ambiguity.
  • Editorial: Because the training markers are 5.5 mm beads and reported errors are of the same scale, any behavior expressed at smaller spatial scales, such as fur ripples or piloerection, is likely invisible to this representation.
  • Editorial: The canonical-surface displacement formulation should transfer to other laboratory animals, such as mice or guinea pigs, provided a small set of shared keypoints and a reference pose can be defined.
  • Editorial: A testable extension is to output a distribution or set of candidate surfaces per keypoint configuration; comparing that multi-modal output on identical-keypoint, different-surface frames would quantify the remaining ambiguity.
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

2 major / 7 minor

Summary. The paper introduces RatDome, a 15-camera multiview dome for capturing freely moving rats wearing colored beads, and RatBodyFormer, a transformer encoder-decoder that maps 10 3D keypoints (nose, eyes, ears, front/back paws, tail base) to a dense set of 3D body-surface points. Training pairs are obtained by triangulating bead and paint markers and mapping them to a canonical surface, with a semi-automatic annotation loop to enlarge the dataset. Experiments on single-rat (D1), two-rat (D2), and markerless all-rat (D3) settings report mean L2 errors of roughly 5.3–7.3 mm on the marker-based splits and 1.8–2.7 mm on D3, together with a large improvement over a MAMMAL-based baseline and an application to an animatable Gaussian splatting model called GaussianRat.

Significance. If the evaluation issues are resolved, this would be a useful first step toward dense surface reconstruction from sparse keypoints for rats, with a novel captured dataset and a network design that explicitly handles inconsistent marker annotations across individuals. The paper has several strengths: the RatDome capture rig and dataset are new assets; the canonical-surface formulation plus point-wise scaling/translation parameters is a sensible way to deal with cross-subject annotations; the ablation studies cover the position consistency loss, the individual-dependent parameters, and data normalization; and Appendix F shows robustness to injected keypoint noise. The main limitations are that the markerless-generalization experiment has a potential leakage between the silhouette loss and the shape-from-silhouette ground truth, and the practical claim of running from 'detected keypoints' is not tested with an actual keypoint detector. These issues currently temper the strength of the central claims but are addressable in a revision.

major comments (2)
  1. [Sec. 4.2 (D3)] In D3, the ground-truth surface points are defined as the sum of LiDAR points and shape-from-silhouette points, and the same SAM-derived mask images are used both to construct the shape-from-silhouette component of that ground truth and to compute the silhouette loss Ls that optimizes the individual-dependent parameters C and T at inference time. Because the shape-from-silhouette target lies on the visual hull of these masks, optimizing Ls can improve agreement with that target for reasons that are not independent of the evaluation metric; this may explain why the reported D3 errors (1.80–2.69 mm) are much smaller than the D1/D2 marker-based errors (5.3–7.3 mm) even though D3 is the harder generalization setting. Please re-evaluate D3 against an independent surface measurement (e.g., marker triangulation on at least a subset of frames), or report errors separately for LiDAR-only and shape-from-silhouette-only points and demonstrate that the reported gains are not an artifact of the shared masks.
  2. [Sec. 4.2 and Appendix F] All quantitative results, including the claimed end-to-end use with 'detected keypoints', are obtained from ground-truth or marker-derived keypoints. Appendix F adds synthetic Gaussian noise to keypoint coordinates but does not run an actual keypoint detector; real detector errors are spatially structured and can differ from isotropic Gaussian noise. Since the abstract and introduction motivate the method by compatibility with past keypoint detectors, please evaluate the full pipeline on videos with an actual 3D keypoint detector (e.g., DANNCE or a DeepLabCut-based detector) and report surface errors conditioned on those detector outputs, or explicitly restrict the claims to keypoint coordinates supplied by other means.
minor comments (7)
  1. [Abstract / Sec. 3.2] The abstract says RatBodyFormer is 'trained with masked-learning', but Sec. 3.2 does not describe any masking procedure; either add the missing definition or remove the term.
  2. [Fig. 1 caption] The caption contains the typo 'Transfomer'; it should read 'Transformer'.
  3. [Appendix B, Eq. (1)] Equation (1) uses both α′ and a′ for the barycentric coordinates; the notation should be unified.
  4. [Sec. 4.1] The paper states that refraction by the acrylic tube is not modeled in the camera calibration; since all marker triangulations pass through this tube, a brief discussion of the expected magnitude of the resulting bias would help readers interpret the reported errors.
  5. [Sec. 6] The conclusion appropriately acknowledges that a single keypoint set may not uniquely determine a body surface in complex interactions; this scope limitation should also be stated in the abstract or introduction so that the headline claim is not broader than the experiments support.
  6. [Sec. 4.2, MAMMAL comparison] The MAMMAL comparison scales a mouse mesh to rat scale but does not retrain or fine-tune MAMMAL on rat data; a brief note on this limitation would help readers interpret the large accuracy gap.
  7. [General] The paper does not state whether the RatDome dataset, trained models, or evaluation code will be released; please add a data-availability statement.

Circularity Check

1 steps flagged · score 4.0 of 10

D3 generalization evaluation is partly circular: shape-from-silhouette ground truth and the test-time silhouette loss both derive from the same SAM masks, while D1/D2 remain independent.

  1. other [Sec. 4.2 (D3: All Rats) paragraph; cf. Sec. 3.2 Loss Functions]
    "For this, we define ground-truth surface points as the sum of the set of points captured by LiDAR and shape-from-silhouette. As the LiDAR points have missing values and errors, we add surface points obtained by shape-from-silhouette. We optimize individual-dependent parameters C and T with the silhouette loss and the position consistency loss for 5 and 14 week-old rats. We obtain the mask images with SAM."

    The D3 evaluation defines its ground-truth surface as LiDAR points plus shape-from-silhouette points derived from SAM masks, and at inference time optimizes the per-rat scaling/translation parameters C and T using the silhouette loss Ls, which penalizes predicted surface points whose projections fall outside the 2D rat region in a view. Because Ls is computed from the same SAM masks used to build the shape-from-silhouette ground truth, the test-time fitting signal and the evaluation target share the same source. The reported D3 errors (1.80 to 2.69 mm) therefore partly measure alignment with the visual hull of those masks rather than an independent surface measurement, so the markerless-generalization claim is not independently established.

full rationale

The core contribution—learning a deterministic mapping from 10 sparse keypoints to a dense set of body-surface points—is trained on RatDome data whose surface labels come from physical beads and paint markers triangulated by multiview geometry, and D1/D2 evaluate on held-out frames against independent manually annotated marker positions. The semi-automatic annotation loop uses RatBodyFormer to assign IDs to triangulated 3D points, but the 3D positions themselves are measured, and the improvement is verified against manual ground truth, so this is standard semi-supervised learning rather than circularity. The one significant circularity concern is the D3 markerless-generalization experiment: the ground-truth surface is defined as LiDAR union shape-from-silhouette from SAM masks, and the same mask-derived silhouette loss is used to fit the per-rat parameters C and T at inference time. The evaluation target and the fitting signal therefore share the same source, which can inflate the reported D3 accuracy; those numbers should not be read as an independent surface measurement. This is a partial, evaluation-level circularity and does not invalidate the D1/D2 results. The Sec. 6 acknowledgment that a single keypoint set may not uniquely determine a body surface point set in complex interactions is a separate limitation and not itself circular.

Assumptions & free parameters 2 free parameters · 3 assumptions · 1 invented entities

The central claim rests on a deterministic keypoint-to-surface mapping, on the correctness of the ARAP-built canonical surface, and on the assumption that the temporary markers do not change behavior. Per-rat scaling and translation parameters are fitted identity parameters, not free constants of a physical model.

free parameters (2)
  • Point-wise scaling parameters C = {CP, CB} for keypoints and body surface points = Optimized per rat during training and at inference; initialized from nose-to-tail length ratio and girth ratio
    These identity-dependent scale factors are fitted to each individual rat in the training data and refined on new rats at test time, and the central claim depends on them.
  • Point-wise translation parameters T = {TP, TB} for keypoints and body surface points = Initialized to 0 and refined during training at every 50 epochs and at inference
    These per-rat translation offsets are fitted identity parameters used in the normalization of the keypoint and surface point displacements.
assumptions (3)
  • domain assumption A single 3D keypoint configuration uniquely determines the 3D body surface for the behavior captured in the dataset.
    RatBodyFormer is a deterministic function from 10 keypoints to N surface points, and the authors acknowledge in Sec. 6 that this does not hold in complex interactions such as multi-rat encounters.
  • domain assumption ARAP deformation of a manually selected reference pose establishes correct semantic correspondences between surface markers of different rats.
    The canonical body surface is built by aligning 7, 9, and 11 week-old rats to a 7-week-old reference using keypoints as hard constraints; no external validation shows that the resulting correspondences are anatomically correct.
  • domain assumption Temporary beads and paint markers are small enough not to alter rat behavior.
    The authors state that markers do not alter behavior as far as they could tell, but no quantitative behavioral comparison is provided.
invented entities (1)
  • Canonical body surface S-tilde
    purpose: Common reference surface used to consolidate marker annotations from different rats and to define the displacement queries and losses for RatBodyFormer.
    The canonical surface is constructed from a manually selected reference pose of one rat and assumed transferable across rats; there is no external validation of anatomical correspondence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RatBodyFormer: Rat Body Surface from Keypoints." pith.science (2026). https://pith.science/paper/O2WFY4DY

@misc{pith2026241209599,
  author       = {Pith},
  title        = {Pith review of: RatBodyFormer: Rat Body Surface from Keypoints},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O2WFY4DY}},
  note         = {Machine review of arXiv:2412.09599}
}
read the original abstract

Analyzing rat behavior lies at the heart of many scientific studies. Past methods for automated rodent modeling have focused on 3D pose estimation from keypoints, e.g., face and appendages. The pose, however, does not capture the rich body surface movement encoding the subtle rat behaviors like curling and stretching. The body surface lacks features that can be visually defined, evading these established keypoint-based methods. In this paper, we introduce the first method for reconstructing the rat body surface as a dense set of points by learning to predict it from the sparse keypoints that can be detected with past methods. Our method consists of two key contributions. The first is RatDome, a novel multi-camera system for rat behavior capture, and a large-scale dataset captured with it that consists of pairs of 3D keypoints and 3D body surface points. The second is RatBodyFormer, a novel network to transform detected keypoints to 3D body surface points. RatBodyFormer is agnostic to the exact locations of the 3D body surface points in the training data and is trained with masked-learning. We experimentally validate our framework with a number of real-world experiments. Our results collectively serve as a novel foundation for automated rat behavior analysis.

Figures

Figures reproduced from arXiv: 2412.09599 by the authors.

Figure 1
Figure 1. We introduce a novel multiview camera system (RatDome) to capture multiview videos of rats (top), and a novel Transfomer [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. A color-beaded rat (left) and RatDome (right). We at [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. RatBodyFormer is an encoder-decoder Transformer model that takes the normalized displacements of detected 3D keypoints and [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Qualitative results of D1. We show the results of trained [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: L2 error histograms of D1. Each vertical bar indicates [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 9
Figure 9. Figure 9: GaussianRat enables photo-realistic rendering of the rat [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: Frames of an animated GaussianRat. eters. Each Gaussian has six properties, location u, scale s, rotation r, opacity σ, color h, and attached mesh face ID k. We define each Gaussian position u by the normal￾ized barycentric coordinates x, y, z and the pose dependent d…
Figure 8
Figure 8. Figure 8: Qualitative comparison between RatBodyFormer (right) [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

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

    cs.CV 2026-07 conditional novelty 6.5 of 10

    A first large multimodal 4D human–dog interaction dataset (6.8M frames) plus an autoregressive model that generates dog motion from human body/hand gestures and audio.

Reference graph

Works this paper leans on

62 extracted references · 59 canonical work pages · cited by 1 Pith paper

  1. [1]

    Cury, Taiga Abe, Venkatesh N

    Alexander Mathis, Pranav Mamidanna, Kevin M. Cury, Taiga Abe, Venkatesh N. Murthy, Mackenzie Weygandt Mathis, and Matthias Bethge. DeepLabCut: markerless pose estimation of user-defined body parts with deep learning.Na- ture Neuroscience, pages 1281—-1289, 2018. 2, 3

    work page 2018

  2. [2]

    Three-dimensional surface motion capture of multiple freely moving pigs using MAMMAL

    Liang An, Jilong Ren, Tao Yu, Tang Hai, Yichang Jia, and Yebin Liu. Three-dimensional surface motion capture of multiple freely moving pigs using MAMMAL. Nature Com- munications, 2023. 2, 3, 6, 7

    work page 2023

  3. [3]

    Monitoring Animal Behavior in the Smart Vivar- ium

    Serge Belongie, Kristin Branson, Piotr Doll ´ar, and Vincent Rabaud. Monitoring Animal Behavior in the Smart Vivar- ium. In Measuring Behavior, pages 70–72, 2005. 3

    work page 2005

  4. [4]

    Whiteway, Cole Hurwitz, Nicholas Greenspan, Robert S

    Dan Biderman, Matthew R. Whiteway, Cole Hurwitz, Nicholas Greenspan, Robert S. Lee, Ankit Vishnub- hotla, Richard Warren, Federico Pedraja, Dillon Noone, Michael M. Schartner, Julia M. Huntenburg, Anup Khanal, Guido T. Meijer, Jean-Paul Noel, Alejandro Pan-Vazquez, Karolina Z. Socha, Anne E. Urai, The International Brain Laboratory, John P. Cunningham, Na...

    work page 2024

  5. [5]

    Federica Bogo, Angjoo Kanazawa, Christoph Lassner, Peter Gehler, Javier Romero, and Michael J. Black. Keep it SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Image. In ECCV, pages 561–578, 2016. 3

    work page 2016

  6. [6]

    Federica Bogo, Javier Romero, Gerard Pons-Moll, and Michael J. Black. Dynamic FAUST: Registering human bod- ies in motion. In CVPR, pages 6233–6242, 2017. 3

    work page 2017

  7. [7]

    ArMo: An Articulated Mesh Approach for Mouse 3D Reconstruction

    James P Bohnslav, Mohammed Abdal Monium Osman, Ak- shay Jaggi, Sofia Soares, Caleb Weinreb, Sandeep Robert Datta, and Christopher D Harvey. ArMo: An Articulated Mesh Approach for Mouse 3D Reconstruction. BioRxiv,

    work page

  8. [8]

    Bola ˜nos, Dongsheng Xiao, Nancy L

    Luis A. Bola ˜nos, Dongsheng Xiao, Nancy L. Ford, Jeff M. LeDue, Pankaj K. Gupta, Carlos Doebeli, Hao Hu, Helge Rhodin, and Timothy H. Murphy. A three-dimensional vir- tual mouse generates synthetic training data for behavioral analysis. Nature Methods, 2021. 2, 3, 7

    work page 2021

Show all 62 references
  1. [9]

    Butlera, Alexander P

    Daniel J. Butlera, Alexander P. Keim, Shantanu Ray, and Eiman Azim. Large-scale capture of hidden fluorescent la- bels for training generalizable markerless motion capture models. Nature Communications, 2023. 3

    2023

  2. [10]

    Computerized video analysis of so- cial interactions in mice

    Fabrice Chaumont, Renata Coura, Pierre Serreau, Arnaud Cressant, Jonathan Chabout, Sylvie Granon, and Jean- Christophe Olivo-Marin. Computerized video analysis of so- cial interactions in mice. Nature Methods, pages 410—-417,

  3. [11]

    Anderson, Kristin Bran- son, Pietro Perona, and Andrew Leifer

    Sandeep Robert Datta, David J. Anderson, Kristin Bran- son, Pietro Perona, and Andrew Leifer. Computational Neu- roethology: A Call to Action. Neuron, pages 11–24, 2019. 1

    2019

  4. [12]

    Dell, John A

    Anthony I. Dell, John A. Bender, Kristin Branson, Iain D. Couzin, Gonzalo G. de Polavieja, Lucas P.J.J. Noldus, Al- fonso P ´erez-Escudero, Pietro Perona, Andrew D. Straw, Martin Wikelski, and Ulrich Brose. Automated image-based tracking and its application in ecology. Trends ...

    2014

  5. [13]

    Dyke, and Juy- ong Zhang

    Bailin Deng, Yuxin Yao, Roberto M. Dyke, and Juy- ong Zhang. A Survey of Non-Rigid 3D Registration. arXiv:2203.07858, 2022. 3

    2022 arXiv

  6. [14]

    Geometric deep learning enables 3D kinematic profiling across species and environments

    Timothy W Dunn, Jesse D Marshall, Kyle S Severson, Diego E Aldarondo, David GC Hildebrand, Selmaan N Chet- tih, William L Wang, Amanda J Gellis, David E Carlson, Dmitriy Aronov, et al. Geometric deep learning enables 3D kinematic profiling across species and environments. Natu...

    2021

  7. [15]

    Roian Egnor and Kristin Branson

    S.E. Roian Egnor and Kristin Branson. Computational Anal- ysis of Behavior. Neuroscience and Biobehavioral Reviews, pages 217–236, 2016. 1

    2016

  8. [16]

    YOLOX: Exceeding YOLO Series in 2021

    Zheng Ge, Songtao Liu, Feng Wang, Zeming Li, and Jian Sun. YOLOX: Exceeding YOLO Series in 2021. arXiv:2107.08430, 2021. 5

    2021 arXiv

  9. [17]

    LiftPose3D, a deep learning-based approach for transforming 2D to 3D pose in laboratory animals

    Adam Gosztolai, Semih G ¨unel, Marco Pietro Abrate, Daniel Morales, Victor R´ıos, Helge Rhodin, Pascal Fua, and Pavan Ramdya. LiftPose3D, a deep learning-based approach for transforming 2D to 3D pose in laboratory animals. Nature Methods, pages 975–981, 2021. 3

    2021

  10. [18]

    Goulding, A

    Evan H. Goulding, A. Katrin Schenk, Punita Juneja, Adri- enne W. MacKay, Jennifer M. Wade, and Laurence H. Tecott. A robust automated system elucidates mouse home cage be- havioral structure. Proceedings of the National Academy of Sciences, pages 20575–20582, 2008. 1

    2008

  11. [19]

    Deep- PoseKit, a software toolkit for fast and robust animal pose estimation using deep learning

    Jacob M Graving, Daniel Chae, Hemal Naik, Liang Li, Ben- jamin Koger, Blair R Costelloe, and Iain D Couzin. Deep- PoseKit, a software toolkit for fast and robust animal pose estimation using deep learning. eLife, 2019. 3

    2019

  12. [20]

    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, pages 7297–7306, 2018. 2, 3

    2018

  13. [21]

    DRWR: A differentiable renderer without render- ing for unsupervised 3D structure learning from silhouette images

    Zhizhong Han, Chao Chen, Yu-Shen Liu, and Matthias Zwicker. DRWR: A differentiable renderer without render- ing for unsupervised 3D structure learning from silhouette images. In ICML, pages 3994–4005, 2020. 5

    2020

  14. [22]

    Multiple view geom- etry in computer vision

    Richard Hartly and Andrew Zisserman. Multiple view geom- etry in computer vision. 2004. 6

    2004

  15. [23]

    Homberg and Markus W ¨ohrand Natalia Alenina

    Judith R. Homberg and Markus W ¨ohrand Natalia Alenina. Comeback of the Rat in Biomedical Research. ACS Chemi- cal Neuroscience, 2017. 2

    2017

  16. [24]

    Tovcimak, J

    Bo Hu, Bryan Seybold, Shan Yang, Avneesh Sud, Yi Liu, Karla Barron, Paulyn Cha, Marcelo Cosino, Ellie Karls- son, Janessa Kite, Ganesh Kolumam, Joseph Preciado, Jos ´e Zavala-Solorio, Chunlian Zhang, Xiaomeng Zhang, Martin V oorbach, Ann E. Tovcimak, J. Graham Ruby, and David ...

    2023

  17. [25]

    Dynamic Multi-Person Mesh Recovery From Uncali- brated Multi-View Cameras

    Buzhen Huang, Yuan Shu, Tianshu Zhang, and Yangang Wang. Dynamic Multi-Person Mesh Recovery From Uncali- brated Multi-View Cameras. In 3DV, pages 710–720. IEEE,

  18. [26]

    Towards accurate marker-less human shape and pose estimation over time

    Yinghao Huang, Federica Bogo, Christoph Lassner, Angjoo Kanazawa, Peter V Gehler, Javier Romero, Ijaz Akhter, and 14 Michael J Black. Towards accurate marker-less human shape and pose estimation over time. In 3DV, pages 421–430,

  19. [27]

    DeeperCut: A Deeper, Stronger, and Faster Multi-Person Pose Estimation Model, 2016

    Eldar Insafutdinov, Leonid Pishchulin, Bjoern Andres, Mykhaylo Andriluka, and Bernt Schiele. DeeperCut: A Deeper, Stronger, and Faster Multi-Person Pose Estimation Model, 2016. 2, 3

    2016

  20. [28]

    Panoptic Studio: A Massively Multiview System for Social Motion Capture

    Hanbyul Joo, Hao Liu, Lei Tan, Lin Gui, Bart Nabbe, Iain Matthews, Takeo Kanade, Shohei Nobuhara, and Yaser Sheikh. Panoptic Studio: A Massively Multiview System for Social Motion Capture. In ICCV, pages 3334–3342, 2015. 4, 5

    2015

  21. [29]

    Ecological validity of social interaction tests in rats and mice

    Kondrakiewicz K, Kostecki M, Szadzi ´nska W, and Knapska E. Ecological validity of social interaction tests in rats and mice. Genes Brain Behav, 2019. 2

    2019

  22. [30]

    Black, David W

    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

  23. [31]

    3D Gaussian Splatting for Real-Time Radiance Field Rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3D Gaussian Splatting for Real-Time Radiance Field Rendering. ACM Transactions on Graphics,

  24. [32]

    Affan, Benjamin B

    Su Jin Kim, Rifqi O. Affan, Benjamin B. Scott Hadas Frostig, and Andrew S. Alexander. Advances in cel- lular resolution microscopy for brain imaging in rats. Neu- rophotonics, 2023. 2

    2023

  25. [33]

    Berg, Wan-Yen Lo, Piotr Dollar, and Ross Girshick

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C. Berg, Wan-Yen Lo, Piotr Dollar, and Ross Girshick. Segment Anything. In ICCV, pages 4015– 4026, 2023. 7

    2023

  26. [34]

    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 ICCV, pages 2252–2261, 2019. 3

    2019

  27. [35]

    The Hungarian method for the assignment problem

    Harold W Kuhn. The Hungarian method for the assignment problem. Naval research logistics quarterly , pages 83–97,

  28. [36]

    Laplace-beltrami eigenfunctions towards an al- gorithm that” understands” geometry

    Bruno L ´evy. Laplace-beltrami eigenfunctions towards an al- gorithm that” understands” geometry. In IEEE International Conference on Shape Modeling and Applications, pages 13– 13, 2006. 5

    2006

  29. [37]

    Tianye Li, Timo Bolkart, Michael. J. Black, Hao Li, and Javier Romero. Learning a model of facial shape and ex- pression from 4D scans. ACM Transactions on Graphics , pages 194:1–194:17, 2017. 3

    2017

  30. [38]

    Matthew Loper, Naureen Mahmood, Javier Romero, Gerard Pons Moll, and Michael J. Black. SMPL: A Skinned Multi- Person Linear Model. ACM Transactions on Graphics, pages 248:1–248:16, 2015. 3, 4

    2015

  31. [39]

    Kyriakou, Ronald Poppe, Els- beth A

    Malte Lorbach, Elisavet I. Kyriakou, Ronald Poppe, Els- beth A. van Dam, Lucas P.J.J. Noldus, and Remco C. Veltkamp. Learning to recognize rat social behavior: Novel dataset and cross-dataset application. Journal of Neuro- science Methods, pages 166–172, 2018. 1

    2018

  32. [40]

    Dynamic 3D Gaussians: Tracking by persis- tent dynamic view synthesis

    Jonathon Luiten, Georgios Kopanas, Bastian Leibe, and Deva Ramanan. Dynamic 3D Gaussians: Tracking by persis- tent dynamic view synthesis. In 3DV, pages 800–809, 2024. 11

    2024

  33. [41]

    Superpixels Based Marker Tracking Vs

    Omid Haji Maghsoudi, Annie Vahedipour Tabrizi, Benjamin Robertson, , and Andrew Spence. Superpixels Based Marker Tracking Vs. Hue Thresholding In Rodent Biomechanics Application. arXiv:1710.06473, 2017. 2, 3

    2017 arXiv

  34. [42]

    Marshall, Diego E

    Jesse D. Marshall, Diego E. Aldarondo, Timothy W. Dunn, William L. Wang, Gordon J. Berman, and Bence P.¨Olveczky. Continuous Whole-Body 3D Kinematic Recordings across the Rodent Behavioral Repertoire. Neuron, pages 420–437,

  35. [43]

    Julieta Martinez, Rayat Hossain, Javier Romero, and James J. Little. A Simple yet Effective Baseline for 3D Hu- man Pose Estimation. In ICCV, pages 2640–2649, 2017. 3

    2017

  36. [44]

    A 3D-Video-Based Comput- erized Analysis of Social and Sexual Interactions in Rats

    Jumpei Matsumoto, Susumu Urakawa, Yusaku Takamura, Renato Malcher-Lopes, Etsuro Hori, Carlos Tomaz, Take- toshi Ono, and Hisao Nishijo. A 3D-Video-Based Comput- erized Analysis of Social and Sexual Interactions in Rats. PLoS One, page e78460, 2013. 2

    2013

  37. [45]

    Dunn, Tuce Tombaz, V

    Bartul Mimica, Benjamin A. Dunn, Tuce Tombaz, V . P. T. N. C. Srikanth Bojja, and Jonathan R. Whitlock. Efficient cor- tical coding of 3D posture in freely behaving rats. Science, pages 584–589, 2018. 2

    2018

  38. [46]

    Davidson, Gabor Vasarhelyi, Daniel Abel, Eniko Kubinyi, Ahmed El Hady, and Tamas Vicsek

    Mate Nagy, Jacob D. Davidson, Gabor Vasarhelyi, Daniel Abel, Eniko Kubinyi, Ahmed El Hady, and Tamas Vicsek. Long-term tracking of social structure in groups of rats. arXiv:2408.08945, 2024. 2

    2024

  39. [47]

    Continu- ous Surface Embeddings

    Natalia Neverova, David Novotny, Marc Szafraniec, Vasil Khalidov, Patrick Labatut, and Andrea Vedaldi. Continu- ous Surface Embeddings. In NeurIPS, pages 17258–17270,

  40. [48]

    T. D. Pereira, D. E. Aldarondo, L. Willmore, M. Kislin, S. S. Wang, M. Murthy, and J. W Shaevitz. Fast animal pose es- timation using deep neural networks. Nature Neuroscience, pages 117–125, 2019. 3

    2019

  41. [49]

    Gaus- sianAvatars: Photorealistic Head Avatars with Rigged 3D Gaussians

    Shenhan Qian, Tobias Kirschstein, Liam Schoneveld, Davide Davoli, Simon Giebenhain, and Matthias Nießner. Gaus- sianAvatars: Photorealistic Head Avatars with Rigged 3D Gaussians. In CVPR, pages 20299–20309, 2024. 8

    2024

  42. [50]

    SAM 2: Segment Anything in Images and Videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman R¨adle, Chloe Rolland, Laura Gustafson, Eric Mintun, Junt- ing Pan, Kalyan Vasudev Alwala, Nicolas Carion, Chao- Yuan Wu, Ross Girshick, Piotr Doll´ar, and Christoph Feic...

    2024 arXiv

  43. [51]

    PSAvatar: A point-based Shape Model for Real-Time Head Avatar Animation with 3D Gaussian Splatting

    Joseph Redmon and Ali Farhadi. PSAvatar: A point-based Shape Model for Real-Time Head Avatar Animation with 3D Gaussian Splatting. arXiv:2401.12900, 2024. 8

    2024 arXiv

  44. [52]

    An- imal Avatars: Reconstructing Animatable 3D Animals from Casual Videos

    Remy Sabathier, Niloy Jyoti Mitra, and David Novotny. An- imal Avatars: Reconstructing Animatable 3D Animals from Casual Videos. arXiv:2403.17103, 2024. 3

    2024 arXiv

  45. [53]

    McCarthy, Andrea Vedaldi, and Natalia Neverova

    Artsiom Sanakoyeu, Vasil Khalidov, Maureen S. McCarthy, Andrea Vedaldi, and Natalia Neverova. Transferring Dense Pose to Proximal Animal Classes. In CVPR, pages 5233– 5242, 2020. 2, 3

    2020

  46. [54]

    15 SplattingAvatar: Realistic Real-Time Human Avatars with Mesh-Embedded Gaussian Splatting

    Zhijing Shao, Zhaolong Wang, Zhuang Li, Duotun Wang, Xiangru Lin, Yu Zhang, Mingming Fan, and Zeyu Wang. 15 SplattingAvatar: Realistic Real-Time Human Avatars with Mesh-Embedded Gaussian Splatting. In CVPR, pages 1606– 1616, 2024. 8

    2024

  47. [55]

    As-Rigid-As-Possible Sur- face Modeling

    Olga Sorkine and Marc Alexa. As-Rigid-As-Possible Sur- face Modeling. In Symposium on Geometry processing , pages 109–116, 2007. 4, 5

    2007

  48. [56]

    Attention is All you Need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszko- reit, Llion Jones, Aidan N Gomez, Ł ukasz Kaiser, and Illia Polosukhin. Attention is All you Need. In NeurIPS, pages 5998–6008, 2017. 4

    2017

  49. [57]

    Superanimal pretrained pose estimation models for behavioral analysis

    Shaokai Ye, Anastasiia Filippova, Jessy Lauer, Steffen Schneider, Maxime Vidal, Tian Qiu, Alexander Mathis, and Mackenzie Weygandt Mathis. Superanimal pretrained pose estimation models for behavioral analysis. Nature Commu- nications, page 5165, 2024. 2, 3, 4

    2024

  50. [58]

    Unified 3D Mesh Recovery of Humans and Animals by Learning Animal Exercise

    Kim Youwang, Kim Ji-Yeon, Kyungdon Joo, and Tae-Hyun Oh. Unified 3D Mesh Recovery of Humans and Animals by Learning Animal Exercise. BMVC, 2021. 3

    2021

  51. [59]

    FreiPose: a deep learning framework for precise animal motion capture in 3D spaces

    Christian Zimmermann, Artur Schneider, Mansour Alyahyay, Thomas Brox, and Ilka Diester. FreiPose: a deep learning framework for precise animal motion capture in 3D spaces. BioRxiv, pages 2020–02, 2020. 2, 3

    2020

  52. [60]

    Automated maternal behavior during early life in rodents (AMBER) pipeline

    Christian Zimmermann, Artur Schneider, Mansour Alyahyay, Thomas Brox, and Ilka Diester. Automated maternal behavior during early life in rodents (AMBER) pipeline. Scientific Reports, 2023. 2, 3

    2023

  53. [61]

    Jacobs, and Michael J

    Silvia Zuffi, Angjoo Kanazawa, David W. Jacobs, and Michael J. Black. 3D Menagerie: Modeling the 3D Shape and Pose of Animals. In CVPR, pages 6365–6373, 2017. 3, 4

    2017

  54. [62]

    Silvia Zuffi, Ylva Mellbin, Ci Li, Markus Hoeschle, Hedvig Kjellstr¨om, Senya Polikovsky, Elin Hernlund, and Michael J. Black. V AREN: Very Accurate and Realistic Equine Net- work. In CVPR, pages 5374–5383, 2024. 3, 4 16

    2024

Pith tools

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