Pith. sign in

REVIEW 4 major objections 7 minor 29 references

Towards Texture- And Shape-Independent 3D Keypoint Estimation in Birds

T0 review · 4 major / 7 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Feeding a pose-estimation network a binary silhouette instead of a textured RGB image preserves 3D multi-pigeon tracking accuracy and lets the pigeon-trained model produce keypoints on other bird species without retraining.

desk verdict A useful but overclaimed extension of 3D-MuPPET: silhouette-only input trades typical keypoint accuracy for fewer extreme outliers, and the 'comparable accuracy' statement only holds under RMSE, not under the paper's own preferred PCK metrics. read the letter →

arxiv 2505.16633 v1 pith:PSCUPXFW submitted 2025-05-22 cs.CV

Valentin Schmuker , Alex Hoi Hang Chan , Bastian Goldluecke , Urs Waldmann This is my paper · ORCID
classification cs.CV
keywords 3Dposeestimationmulti-animaltrackingsilhouette-basedkeypointstextureindependencespeciestransfersegmentationmasksmulti-viewtriangulationpigeons
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to show that a bird's silhouette carries enough information to estimate and track 3D joint positions, so that texture is not needed. It replaces the textured image input of an existing multi-pigeon pose-and-tracking framework with binary masks produced by a segmentation model, then retrains the 2D keypoint network on those masks. On the 3D-POP benchmark the silhouette-only pipeline reaches accuracy comparable to the texture-based original, and on a harder outdoor dataset it reports a lower average 3D error. The same pigeon-trained model yields preliminary 2D keypoint estimates on four other bird species without fine-tuning. If this holds, researchers could track animal pose in new species and environments without collecting species-specific textured training data.

What carries the argument

The central object is the binary silhouette: a segmentation mask reduced to its largest connected component, an object the paper calls DLCISO. This object does the work of stripping away texture, color, and illumination while preserving the shape cues the keypoint network needs. The mask is generated by prompting a segmentation model with the detector's bounding box, and it is the only input seen by the pose network. The paper treats this abstraction as the mechanism that makes the framework both texture-independent and transferable across species.

What would settle it

Take frames in which two pigeons occlude each other so their masks merge into one connected component, and compare per-keypoint accuracy on those frames against accuracy on frames with isolated birds; if median keypoint error jumps and the fraction of keypoints within the strict accuracy threshold collapses specifically in merged-mask frames while the textured-input baseline does not, the assumption that the largest component equals the bird is falsified.

Watch

Extended reading notes

Core claim

The central claim is that texture is not a necessary input for multi-animal 3D keypoint estimation in birds. By converting each detected bird into a binary silhouette and training a keypoint regressor on those silhouettes, the authors preserve the original framework's ability to estimate and track 2D and 3D poses of up to ten pigeons from four camera views. They report that this texture-independent version achieves comparable accuracy to the texture-dependent baseline, and on the Wild-MuPPET outdoor dataset it attains a lower average 3D error (32.1 mm versus 53.4 mm) while retaining a higher median error (21.5 mm versus 15.0 mm). The pigeon-trained silhouette model also produces 2D keypoints on pigeons, stock doves, turtle doves, and ravens without fine-tuning, with best results when the new species' pose and viewing angle resemble the training data.

Load-bearing premise

The single largest connected component of the segmentation mask is assumed to be the complete bird, so whenever birds overlap, occlude one another, or split into fragments, the silhouette no longer contains enough information to localize all nine keypoints.

Editorial extensions

If this is right

  • Silhouette-only input is enough to keep 2D and 3D pose estimation and identity tracking working for up to 10 pigeons across four camera views.
  • Because the input is a mask, the pipeline is insensitive to plumage color, lighting, and background changes that leave the silhouette unchanged.
  • A model trained only on pigeon silhouettes can generate usable 2D keypoints on other bird species without additional labeled data.
  • On the outdoor Wild-MuPPET dataset, the silhouette variant with segmentation-model masks reports lower average 3D error than the texture-dependent baseline, indicating fewer large outliers.

Reading between the lines

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

  • The paper's logic suggests that synthetic silhouette data, such as rendered bird shapes with random poses, could be used to train the keypoint network and remove the annotation bottleneck entirely.
  • A testable extension: triangulating all four left/right keypoint combinations and keeping the one with the smallest reprojection error should resolve the left-right ambiguity the paper identifies in silhouette data.
  • Because the authors find mask quality is the main bottleneck, adding internal contour cues to the silhouette might push accuracy beyond the texture-based baseline rather than merely matching it.
  • The species-transfer results imply the real barrier to new species is pose and viewpoint distribution shift, not feather appearance, so viewpoint augmentation or a few annotated frames of the new species may close most of the gap.
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 / 7 minor

Summary. The paper proposes a texture-independent extension of the 3D-MuPPET framework for multi-pigeon 2D/3D pose estimation and tracking. Instead of feeding RGB crops to DeepLabCut, the method generates silhouettes via SAM (DLCSAM) or via SAM with largest-connected-component post-processing (DLCISO), and on Wild-MuPPET also via YOLOv8-seg (DLCISO-YOLO). The 2D keypoints are triangulated to 3D and tracked as in 3D-MuPPET. The authors report results on the 3D-POP and Wild-MuPPET benchmarks, an ablation of SAM variants, a keypoint-wise error analysis, and zero-shot species-transfer experiments on four bird species from Animal Kingdom. The central claim, stated in the abstract and Sec. 6, is that the texture-independent approach achieves accuracy comparable to the original texture-dependent 3D-MuPPET while being robust to texture and illumination changes.

Significance. If the central claim were fully supported, the paper would make a useful contribution: silhouette-only input would enable pose estimation to transfer across appearances and to new species without fine-tuning, which is valuable for bird behavior research. The paper has several strengths: it evaluates on held-out public benchmarks (3D-POP and Wild-MuPPET), uses zero-shot species transfer rather than retraining, includes a keypoint-level error analysis, and reports ablation studies on SAM variants and data augmentation. No circularity issue is present: the method is trained on standard annotations and evaluated on held-out splits, and the comparison with 3D-MuPPET uses the same splits. However, the main claim of "comparable accuracy" is not supported by the paper's own preferred metrics (PCK and median error), and several auxiliary claims, such as "preliminary promising results" for species transfer, are stated without quantitative baselines or thresholds. The paper is therefore a reasonable proof-of-concept but needs substantive revision of its claims and evidence before publication.

major comments (4)
  1. [Sec. 4.1, Sec. 4.2, Tabs. 2 and 3] The central claim of "comparable accuracy" rests solely on RMSE. Sec. 4.1 states that PCK is "the more meaningful performance measure," and on every such measure the silhouette methods are substantially worse: on 3D-POP (Tab. 2), median error rises from 7.5 mm to 13.5 mm, PCK05 drops from 66.1% to 35.4%, and PCK10 drops from 90.9% to 74.5%; on Wild-MuPPET (Tab. 3), median rises from 15.0 mm to 21.5 mm, PCK05 drops from 25.1% to 8.8%, and PCK10 drops from 74.4% to 48.3%. A lower RMSE combined with a higher median indicates a distribution that is more concentrated around a less accurate center, not a generally comparable predictor. The authors interpret the lower RMSE as a reduction in outliers, but they provide no outlier definition, no error CDF or quantile breakdown, and no task-level analysis showing that the RMSE improvement matters for downstream use. The "comparable accuracy" claim is therefore not established by the reported evidence and should be replaced with a claim that accurately reflects the metric trade-off, or supported with the missing distributional analysis.
  2. [Sec. 3.2, Sec. 5] The largest-connected-component assumption is load-bearing for the DLCISO variant and is acknowledged in Sec. 5 only as a general limitation. In multi-animal scenes with up to 10 pigeons, overlapping birds can split into disconnected mask fragments or merge into a single component, and the assumption that the largest component is the target pigeon has no direct validation. The paper reports aggregate results over all group sizes but no breakdown by number of animals, occlusion level, or mask quality, so the reader cannot tell when the assumption fails and how often it fails. This should be quantified or at least analyzed on a subset of frames, because the multi-animal claim is part of the paper's stated contribution.
  3. [Sec. 4.4, Tab. 6] The species-transfer results are presented as "preliminary promising results," but the only numerical evidence is Table 6, where PCK10 ranges from 19.9% to 51.9% and PCK05 from 8.5% to 14.0%. Without a baseline (e.g., the texture-dependent model under the same protocol, or a chance-level bound) or a task-level demonstration of usefulness, these numbers do not by themselves establish that the model "is capable of generalizing to new species." Additionally, the 73 raven images were manually selected for similar angle and behavior to 3D-POP; the selection criteria are subjective and not described in enough detail to judge whether the reported PCK10 of 51.9% is representative. The paper should either temper the conclusion or provide a more rigorous evaluation, including the number of images per species and a comparison against a suitable baseline.
  4. [Sec. 4.3, Tab. 4] The keypoint-wise analysis is informative but also reveals that the method is not yet reliable for a major body part: the tail has a median error of 21.0 mm and PCK05 of 1.9%, while left/right shoulder and eye confusion is admitted in Sec. 5. These are not merely minor imperfections; they directly affect the usability of the estimated pose for behavioral analysis. The manuscript should either discuss the consequences of these errors for the intended downstream use or restrict the claims to keypoints that can be localized reliably from silhouettes.
minor comments (7)
  1. [Throughout] There are several typos and inconsistencies: "Waldman et al." should be "Waldmann et al."; "tine-tuning" in Sec. 1 should be "fine-tuning"; "apperance" in Sec. 3.2 should be "appearance"; and the table captions use "strived for" rather than the more standard "lower is better / higher is better."
  2. [Sec. 4.1] The definition of the PCK threshold in 3D is given as "the maximum distance between any two keypoints," but it is not stated whether this is computed per frame from ground-truth keypoints and whether it matches the protocol used by 3D-MuPPET. Please clarify for reproducibility.
  3. [Sec. 4.2] Tables 2 and 3 report single aggregate values per metric, but the test set for 3D-POP consists of four sequences and Wild-MuPPET is a single pigeon; no variance or per-sequence breakdown is given. Please report per-sequence results or confidence intervals, especially given the small number of test sequences.
  4. [Sec. 4.3, Tab. 5] The SAM ablation reports only three configurations (huge, base, base fine-tuned). It would be helpful to state the number of fine-tuning iterations and the fine-tuning data source, since the conclusion that fine-tuning has "limited benefit" depends on this experimental setup.
  5. [Sec. 4.4] The species-transfer section does not report the number of images evaluated per species, and the text says "we evaluate on all images" without giving the count. Please include sample sizes and, if possible, the keypoint-level breakdown for the transfer results.
  6. [Sec. 5] The limitation that silhouettes make left-right keypoint assignment ambiguous is important, but the proposed solution (triangulating different left-right combinations and picking the smallest triangulation error) is not tested. Consider adding a small experiment or at least a more concrete description of how this would be implemented.
  7. [Sec. 6] The conclusion repeats the "comparable accuracy" claim without the caveat that it applies only to RMSE and not to the paper's own preferred metrics. The conclusion should be revised to be consistent with the metric analysis requested in the major comments.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the pipeline is trained on one dataset and evaluated on held-out external benchmarks, with the baseline comparison against prior work being a fair benchmark rather than a self-referential loop.

full rationale

The paper's core empirical claim is that a silhouette-based pose estimator, trained from scratch on 3D-POP silhouettes, achieves accuracy comparable to the texture-dependent 3D-MuPPET baseline. The training data, validation criterion, and test splits are external to the paper's own construction: the authors state they 'use the same training, validation and test data as 3D-MuPPET [27]' and report metrics on held-out 3D-POP and Wild-MuPPET sequences. The comparison against 3D-MuPPET is a baseline comparison, not a derivation: the cited prior framework supplies the triangulation, matching, and tracking machinery, but the keypoint predictor is independently trained on silhouettes and evaluated on ground truth annotations. No parameter is fitted to the test set and then renamed a prediction; no equation defines the output in terms of the target; and no uniqueness theorem or self-citation is invoked to forbid alternative choices. The skeptical concern that 'comparable accuracy' rests only on RMSE while the paper's own preferred PCK metrics show degradation is a statistical-evidence critique, not a circularity critique: it concerns how strongly the reported numbers support the claim, not whether the claim reduces to its inputs. Similarly, the acknowledged limitation that the largest-connected-component assumption fails under occlusion and that tail keypoints are poorly localized is an honest statement of failure modes, not a circular step. The species-transfer experiments are zero-shot, using a model trained on pigeon silhouettes and applied to Animal Kingdom images without tuning, which is an externally falsifiable test. There is no circular step in the derivation chain, so the appropriate finding is no significant circularity.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

No invented entities are introduced. The claims rest on domain assumptions about silhouette informativeness and mask quality, plus standard multi-view geometry. There are no fitted constants in the sense of a derivation; the listed items are design choices and assumptions inherited from the method.

free parameters (3)
  • Largest-connected-area mask selection
    Ad hoc heuristic in DLCISO (Sec. 3.2) that picks the largest continuous mask region as the target pigeon; chosen by hand, not learned, and can fail under occlusion or fragmentation.
  • Augmentation configuration for DLC retraining = disabled contrast/noise/blur, enabled left-right flip
    Chosen by hand in Sec. 4.3 ablation; affects model performance (RMSE 21.1 vs 22.2 mm), showing sensitivity to training choices.
  • SAM model variant = SAM-huge
    Selected among huge, base, and fine-tuned base in Sec. 4.3; SAM-huge gave best median and PCK10, but fine-tuned base gave lower RMSE. Model choice is a free design decision.
assumptions (4)
  • domain assumption Silhouettes contain sufficient information to recover 2D keypoints
    The entire extension rests on this; Sec. 5 admits left-right ambiguity and poor tail accuracy as consequences.
  • ad hoc to paper The largest continuous region of the SAM mask is the target pigeon
    Stated in Sec. 3.2 for DLCISO; fails when birds overlap or the mask is fragmented.
  • domain assumption YOLOv8 bounding-box detections are texture-invariant 'bird' detections
    Assumed in Sec. 3.2 because YOLOv8 is trained on a multi-species dataset, so the same weights are used as in 3D-MuPPET.
  • standard math Multi-view triangulation of 2D keypoints yields valid 3D keypoints
    Standard multi-view geometry, used identically to 3D-MuPPET.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Texture- And Shape-Independent 3D Keypoint Estimation in Birds." pith.science (2026). https://pith.science/paper/PSCUPXFW

@misc{pith2026250516633,
  author       = {Pith},
  title        = {Pith review of: Towards Texture- And Shape-Independent 3D Keypoint Estimation in Birds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PSCUPXFW}},
  note         = {Machine review of arXiv:2505.16633}
}
read the original abstract

In this paper, we present a texture-independent approach to estimate and track 3D joint positions of multiple pigeons. For this purpose, we build upon the existing 3D-MuPPET framework, which estimates and tracks the 3D poses of up to 10 pigeons using a multi-view camera setup. We extend this framework by using a segmentation method that generates silhouettes of the individuals, which are then used to estimate 2D keypoints. Following 3D-MuPPET, these 2D keypoints are triangulated to infer 3D poses, and identities are matched in the first frame and tracked in 2D across subsequent frames. Our proposed texture-independent approach achieves comparable accuracy to the original texture-dependent 3D-MuPPET framework. Additionally, we explore our approach's applicability to other bird species. To do that, we infer the 2D joint positions of four bird species without additional fine-tuning the model trained on pigeons and obtain preliminary promising results. Thus, we think that our approach serves as a solid foundation and inspires the development of more robust and accurate texture-independent pose estimation frameworks.

Figures

Figures reproduced from arXiv: 2505.16633 by the authors.

Figure 1
Figure 1. Qualitative results of bird poses using our texture￾independent approach. Top: Multi-pigeon pose estimation and tracking in 3D, projected to 2D. Green lines connect the body, blue lines the head keypoints. The detected bounding boxes are red. Example frame from 3D-POP [20]. Bottom left and right: Species transfer to turtle doves and ravens, respectively. Detected bounding box in red. Both examples from [21]. However… view at source ↗
Figure 2
Figure 2. Texture independence. Two examples of our mask gen￾eration from textured images (left) using DLCSAM (middle) and DLCISO (right), cf. Sec. 3.2. Example frames from [20]. ground truth bounding boxes are provided we detect them with YOLOv8 [25]. 3.2. 3D-MuPPET 3D-MuPPET [27] is a 3D multi-pigeon pose estimation and tracking framework that consists of a pose estimation and a tracking module. Waldman et al. [27] use thre… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 29 canonical work pages

  1. [1]

    Pfrommer, Marc F

    Marc Badger, Yufu Wang, Adarsh Modh, Ammon Perkes, Nikos Kolotouros, Bernd G. Pfrommer, Marc F. Schmidt, and Kostas Daniilidis. 3D bird reconstruction: A dataset, model, and shape recovery from a single view. InComputer Vision – ECCV 2020, pages 1–17. Springer Nature, 2020. 2, 3

    work page 2020

  2. [2]

    Simple online and realtime tracking

    Alex Bewley, Zongyuan Ge, Lionel Ott, Fabio Ramos, and Ben Upcroft. Simple online and realtime tracking. In2016 IEEE International Conference on Image Processing (ICIP), pages 3464–3468, 2016. 3

    work page 2016

  3. [3]

    Creatures great and SMAL: Recovering the shape and motion of animals from video

    Benjamin Biggs, Thomas Roddick, Andrew Fitzgibbon, and Roberto Cipolla. Creatures great and SMAL: Recovering the shape and motion of animals from video. InComputer Vision - ACCV 2018, pages 3–19. Springer Nature, 2019. 2

    work page 2018

  4. [4]

    Peering into the world of wild passer- ines with 3d-socs: synchronized video capture and posture estimation.bioRxiv, pages 2024–06, 2024

    Michael Chimento, Alex Hoi Hang Chan, Lucy M Aplin, and Fumihiro Kano. Peering into the world of wild passer- ines with 3d-socs: synchronized video capture and posture estimation.bioRxiv, pages 2024–06, 2024. 1, 2

    work page 2024

  5. [5]

    Couzin and Conor Heins

    Iain D. Couzin and Conor Heins. Emerging technologies for behavioral research in changing environments.Trends in Ecology and Evolution, 38(4):346–354, 2023. 1

    work page 2023

  6. [6]

    Estimation of human body 3D pose for parent-infant interaction settings using azure Kinect and OpenPose.MethodsX, 13:102861,

    Franc ¸oise Diaz-Rojas and Masako Myowa. Estimation of human body 3D pose for parent-infant interaction settings using azure Kinect and OpenPose.MethodsX, 13:102861,

    work page

  7. [7]

    Dunn, Jesse D

    Timothy W. Dunn, Jesse D. Marshall, Kyle S. Severson, Diego E. Aldarondo, David G. C. Hildebrand, Selmaan N. Chettih, William L. Wang, Amanda J. Gellis, David E. Carl- son, Dmitriy Aronov, Winrich A. Freiwald, Fan Wang, and Bence P. ¨Olveczky. Geometric deep learning enables 3D kinematic profiling across species and environments.Nature Methods, 18(5):564–...

    work page 2021

  8. [8]

    Neural puppeteer: Keypoint-based neural rendering of dynamic shapes

    Simon Giebenhain, Urs Waldmann, Ole Johannsen, and Bas- tian Goldluecke. Neural puppeteer: Keypoint-based neural rendering of dynamic shapes. InProceedings of the Asian Conference on Computer Vision (ACCV), pages 2830–2847,

    work page

Show all 29 references
  1. [9]

    Multi-animal 3d social pose estimation, identification and behaviour embedding with a few-shot learning framework.Nature Machine Intelligence, 6(1):48–61, 2024

    Yaning Han, Ke Chen, Yunke Wang, Wenhao Liu, Zhouwei Wang, Xiaojing Wang, Chuanliang Han, Jiahui Liao, Kang Huang, Shengyuan Cai, et al. Multi-animal 3d social pose estimation, identification and behaviour embedding with a few-shot learning framework.Nature Machine Intelligenc...

    2024

  2. [10]

    Congzhentao Huang, Shuai Jiang, Yang Li, Ziyue Zhang, Jason Traish, Chen Deng, Sam Ferguson, and Richard Y . Da Xu. End-to-end dynamic matching network for multi- view multi-person 3D pose estimation. InComputer Vision - ECCV 2020, pages 477–493. Springer Nature, 2020. 3

    2020

  3. [11]

    Islam, Jiawei Mo, and Junaed Sattar

    Md J. Islam, Jiawei Mo, and Junaed Sattar. Robot-to-robot relative pose estimation using humans as markers.Au- tonomous robots, 45(4):579–593, 2021. 1

    2021

  4. [12]

    Acinoset: a 3d pose estimation dataset and baseline models for cheetahs in the wild

    Daniel Joska, Liam Clark, Naoya Muramatsu, Ricardo Jericevich, Fred Nicolls, Alexander Mathis, Mackenzie W Mathis, and Amir Patel. Acinoset: a 3d pose estimation dataset and baseline models for cheetahs in the wild. In2021 IEEE international conference on robotics and automati...

    2021

  5. [13]

    Deci- phering social traits and pathophysiological conditions from natural behaviors in common marmosets.Current Biology, 34(13):2854–2867, 2024

    Takaaki Kaneko, Jumpei Matsumoto, Wanyi Lu, Xincheng Zhao, Louie Richard Ueno-Nigh, Takao Oishi, Kei Kimura, Yukiko Otsuka, Andi Zheng, Kensuke Ikenaka, et al. Deci- phering social traits and pathophysiological conditions from natural behaviors in common marmosets.Current Biol...

    2024

  6. [14]

    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. InProceedings of the IEEE/CVF International Conference on Computer Vision...

    2023

  7. [15]

    Akoad, Bence P

    Ugne Klibaite, Tianqing Li, Diego Aldarondo, Jumana F. Akoad, Bence P. ¨Olveczky, and Timothy W. Dunn. Mapping the landscape of social behavior.Cell, 0(0), 2025. Publisher: Elsevier. 2

    2025

  8. [16]

    Kerby, Ja- cob M

    Benjamin Koger, Adwait Deshpande, Jeffrey T. Kerby, Ja- cob M. Graving, Blair R. Costelloe, and Iain D. Couzin. Quantifying the movement, behaviour and environmental context of group-living animals using drones and computer vision.Journal of Animal Ecology, 92(7):1357–1371, 2023. 1

    2023

  9. [17]

    Rah- man, Valentina Di Santo, Daniel Soberanes, Guoping Feng, Venkatesh N

    Jessy Lauer, Mu Zhou, Shaokai Ye, William Menegas, Steffen Schneider, Tanmay Nath, Mohammed M. Rah- man, Valentina Di Santo, Daniel Soberanes, Guoping Feng, Venkatesh N. Murthy, George Lauder, Catherine Dulac, Mackenzie W. Mathis, and Alexander Mathis. Multi-animal pose estima...

    2022

  10. [18]

    Cury, Taiga Abe, Venkatesh N

    Alexander Mathis, Pranav Mamidanna, Kevin M. Cury, Taiga Abe, Venkatesh N. Murthy, Mackenzie W. Mathis, and Matthias Bethge. DeepLabCut: Markerless pose estimation of user-defined body parts with deep learning.Nature Neu- roscience, 21(9):1281–1289, 2018. 1, 2, 3

    2018

  11. [19]

    Smart- barn: Scalable multimodal arena for real-time tracking be- havior of animals in large numbers.Science Advances, 9 (35):eadf8068, 2023

    M ´at´e Nagy, Hemal Naik, Fumihiro Kano, Nora V Carlson, Jens C Koblitz, Martin Wikelski, and Iain D Couzin. Smart- barn: Scalable multimodal arena for real-time tracking be- havior of animals in large numbers.Science Advances, 9 (35):eadf8068, 2023. 1

    2023

  12. [20]

    Couzin, Fumihiro Kano, and M ´at´e Nagy

    Hemal Naik, Alex Hoi Hang Chan, Junran Yang, Mathilde Delacoux, Iain D. Couzin, Fumihiro Kano, and M ´at´e Nagy. 3D-POP - an automated annotation approach to fa- cilitate markerless 2D-3D tracking of freely moving birds with marker-based motion capture. InProceedings of the IE...

    2023

  13. [21]

    Animal kingdom: A large and diverse dataset for animal behavior understanding

    Xun Long Ng, Kian Eng Ong, Qichen Zheng, Yun Ni, Si Yong Yeo, and Jun Liu. Animal kingdom: A large and diverse dataset for animal behavior understanding. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 19023–19034, 2022. 1, 2, 4, 5

    2022

  14. [22]

    Pereira, Diego E

    Talmo D. Pereira, Diego E. Aldarondo, Lindsay Willmore, Mikhail Kislin, Samuel S. H. Wang, Mala Murthy, and 6 Joshua W. Shaevitz. Fast animal pose estimation using deep neural networks.Nature Methods, 16(1):117–125, 2019. 2

    2019

  15. [23]

    Pereira, Nathaniel Tabris, Arie Matsliah, David M

    Talmo D. Pereira, Nathaniel Tabris, Arie Matsliah, David M. Turner, Junyu Li, Shruthi Ravindranath, Eleni S. Papadoyan- nis, Edna Normand, David S. Deutsch, Z. Y . Wang, Grace C. McKenzie-Smith, Catalin C. Mitelut, Marielisa D. Castro, John D’Uva, Mikhail Kislin, Dan H. Sanes,...

    2022

  16. [24]

    UniAP: Towards Universal Animal Perception in vi- sion via few-shot learning

    Meiqi Sun, Zhonghan Zhao, Wenhao Chai, Hanjun Luo, Shi- dong Cao, Yanting Zhang, Jenq-Neng Hwang, and Gaoang Wang. UniAP: Towards Universal Animal Perception in vi- sion via few-shot learning. InProceedings of the AAAI Con- ference on Artificial Intelligence, pages 5008–5016. ...

    2024

  17. [25]

    YOLOv8: A novel object detection algorithm with enhanced performance and robust- ness

    Rejin Varghese and Sambath M. YOLOv8: A novel object detection algorithm with enhanced performance and robust- ness. In2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems (ADICS), pages 1–6, 2024. 3

    2024

  18. [26]

    Improving unsupervised label propagation for pose tracking and video object segmen- tation

    Urs Waldmann, Jannik Bamberger, Ole Johannsen, Oliver Deussen, and Bastian Goldl ¨ucke. Improving unsupervised label propagation for pose tracking and video object segmen- tation. InDAGM German Conference on Pattern Recogni- tion, pages 230–245, 2022. 1

    2022

  19. [27]

    Urs Waldmann, Alex H. H. Chan, Hemal Naik, M ´at´e Nagy, Iain D. Couzin, Oliver Deussen, Bastian Goldluecke, and Fu- mihiro Kano. 3D-MuPPET: 3D multi-pigeon pose estima- tion and tracking.International Journal of Computer Vision, 132(10):4235–4252, 2024. 1, 2, 3, 4, 5

    2024

  20. [28]

    ViTPose: Simple vision transformer baselines for human pose estimation

    Yufei Xu, Jing Zhang, Qiming Zhang, and Dacheng Tao. ViTPose: Simple vision transformer baselines for human pose estimation. InAdvances in Neural Information Process- ing Systems, pages 38571–38584. Curran Associates, Inc.,

  21. [29]

    Superanimal pretrained pose estimation models for behavioral analysis.Nature Commu- nications, 15(1), 2024

    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, 15(1), 2024. 1, 2 7

    2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.