FREE SHIPPING over $40
OVER 9000 FIVE-STAR REVIEWS
5% OFF ALL SUBSCRIPTIONS
100% HAPPINESS GUARANTEE
Analyze Diet
0
Home / Equine Research Bank / Sci Data / The Poses for Equine Research Dataset (PFERD).
Scientific data2024; 11(1); 497; doi: 10.1038/s41597-024-03312-1

The Poses for Equine Research Dataset (PFERD).

Abstract: Studies of quadruped animal motion help us to identify diseases, understand behavior and unravel the mechanics behind gaits in animals. The horse is likely the best-studied animal in this aspect, but data capture is challenging and time-consuming. Computer vision techniques improve animal motion extraction, but the development relies on reference datasets, which are scarce, not open-access and often provide data from only a few anatomical landmarks. Addressing this data gap, we introduce PFERD, a video and 3D marker motion dataset from horses using a full-body set-up of densely placed over 100 skin-attached markers and synchronized videos from ten camera angles. Five horses of diverse conformations provide data for various motions from basic poses (eg. walking, trotting) to advanced motions (eg. rearing, kicking). We further express the 3D motions with current techniques and a 3D parameterized model, the hSMAL model, establishing a baseline for 3D horse markerless motion capture. PFERD enables advanced biomechanical studies and provides a resource of ground truth data for the methodological development of markerless motion capture.
© 2024. The Author(s).
Get Full Text
Publication Date: 2024-05-15 PubMed ID: 38750064PubMed Central: PMC11096353DOI: 10.1038/s41597-024-03312-1Google Scholar: Lookup
The Equine Research Bank provides access to a large database of publicly available scientific literature. Inclusion in the Research Bank does not imply endorsement of study methods or findings by Mad Barn.
LinkedInCopy
  • Dataset
  • Journal Article
  • Research Support
  • Non-U.S. Gov't

Summary

This research summary has been generated with artificial intelligence and may contain errors and omissions. Refer to the original study to confirm details provided. Submit correction.

Overview

  • This study presents PFERD, a comprehensive dataset capturing detailed 3D motions of horses using over 100 skin-attached markers and synchronized multi-camera videos.
  • The dataset aims to support advanced biomechanical research and the development of markerless motion capture techniques for quadruped animals, especially horses.

Introduction and Context

  • Understanding quadruped animal motion is crucial for diagnosing diseases, analyzing animal behavior, and studying the physical mechanics behind various gaits.
  • Horses are among the most researched quadrupeds due to their importance in biomechanics, sports, and veterinary science.
  • Traditional data capture methods using physical markers on animals are difficult, time-consuming, and often limited to a few key anatomical points.
  • Computer vision offers promising alternatives for motion capture, but its advancement requires high-quality, detailed reference datasets that are currently rare and usually not openly accessible.

About the PFERD Dataset

  • PFERD stands for Poses for Equine Research Dataset, designed to fill the data availability gap for horse motion studies.
  • The dataset includes:
    • 3D motion data collected through a dense arrangement of over 100 skin-attached markers covering the entire body of the horse.
    • Video recordings synchronized from ten different camera angles, enabling multi-view analysis.
    • Create a comprehensive range of motions captured—from basic locomotion like walking and trotting to complex actions such as rearing and kicking.
  • The dataset involves five horses of varying body types and conformations, enhancing the dataset’s generalizability and diversity.

Technical Contributions

  • The researchers applied state-of-the-art motion analysis techniques to describe the collected 3D motion data.
  • They employed a 3D parametric horse model named hSMAL (a specialized model designed to represent the horse’s shape and motion) to express and interpret the 3D motions effectively.
  • This modeling approach establishes a baseline for developing and benchmarking new 3D markerless motion capture algorithms tailored specifically to horses.

Significance and Applications

  • PFERD offers a groundbreaking resource for the biomechanical study of equine movement, allowing researchers to:
    • Analyze detailed full-body motion patterns with high precision.
    • Investigate locomotion dynamics and pathological gait abnormalities with more depth and accuracy than before.
    • Develop and validate computer vision models that do not rely on physical markers, facilitating non-invasive markerless motion capture systems.
  • The open-access nature of PFERD promotes collaborative research and accelerates innovation in animal motion capture methodologies.
  • It encourages advancements that can benefit veterinary diagnostics, animal welfare monitoring, sports science, and broader biological studies involving horses and potentially other quadrupeds.

Summary

  • The PFERD dataset addresses a vital data scarcity in equine motion research by providing dense 3D motion capture data along with synchronized multi-view videos.
  • Its comprehensive scope, covering multiple horses and a wide range of motions, combined with advanced motion representation using the hSMAL model, sets a new standard for research and development in equine biomechanics and computer vision-based motion capture.

Cite This Article

APA
Li C, Mellbin Y, Krogager J, Polikovsky S, Holmberg M, Ghorbani N, Black MJ, Kjellström H, Zuffi S, Hernlund E. (2024). The Poses for Equine Research Dataset (PFERD). Sci Data, 11(1), 497. https://doi.org/10.1038/s41597-024-03312-1

Publication

Scientific data
ISSN: 2052-4463
NlmUniqueID: 101640192
Country: England
Language: English
Volume: 11
Issue: 1
Pages: 497
PII: 497

Researcher Affiliations

Li, Ci
  • KTH Royal Institute of Technology, Stockholm, Sweden.
Mellbin, Ylva
  • Swedish University of Agricultural Sciences, Uppsala, Sweden.
Krogager, Johanna
  • Swedish University of Agricultural Sciences, Uppsala, Sweden.
Polikovsky, Senya
  • Max Planck Institute for Intelligent Systems, Tübingen, Germany.
Holmberg, Martin
  • Qualisys, Göteborg, Sweden.
Ghorbani, Nima
  • Sporttotal.tv, Immersive Technologies, Cologne, Germany.
Black, Michael J
  • Max Planck Institute for Intelligent Systems, Tübingen, Germany.
Kjellström, Hedvig
  • KTH Royal Institute of Technology, Stockholm, Sweden.
  • Swedish University of Agricultural Sciences, Uppsala, Sweden.
Zuffi, Silvia
  • CNR Institute for Applied Mathematics and Information Technologies, Milan, Italy.
Hernlund, Elin
  • Swedish University of Agricultural Sciences, Uppsala, Sweden. elin.hernlund@slu.se.

MeSH Terms

  • Animals
  • Biomechanical Phenomena
  • Gait
  • Horses / physiology
  • Video Recording

Conflict of Interest Statement

The authors declare no competing interests.

References

This article includes 55 references
  1. Louis N. Temporally guided articulated hand pose tracking in surgical videos. International Journal of Computer Assisted Radiology and Surgery 2023;18:117–125.
    doi: 10.1007/s11548-022-02761-6pmc: PMC9883342pubmed: 36190616google scholar: lookup
  2. Zhang JZ. Slomo: A general system for legged robot motion imitation from casual videos. 7154–7161, 2023.
    doi: 10.1109/lra.2023.3313937google scholar: lookup
  3. Luo H. Artemis: Articulated neural pets with appearance and motion synthesis. 2022.
    doi: 10.1145/3528223.3530086google scholar: lookup
  4. Khokhlova M, Migniot C, Morozov A, Sushkova O, Dipanda A. Normal and pathological gait classification lstm model. Artificial Intelligence in Medicine 2019;94:54–66.
    doi: 10.1016/j.artmed.2018.12.007pubmed: 30871683google scholar: lookup
  5. Raab S. Single motion diffusion. 2024.
    doi: 10.48550/arxiv.2302.05905google scholar: lookup
  6. Guo C. Generating diverse and natural 3d human motions from text. 5152–5161, 2022.
    doi: 10.1109/cvpr52688.2022.00509google scholar: lookup
  7. Mir A, Puig X, Kanazawa A, Pons-Moll G. Generating continual human motion in diverse 3d scenes. 2024.
    doi: 10.48550/arxiv.2304.02061google scholar: lookup
  8. Wilson A, Weller R. The biomechanics of the equine limb and its effect on lameness. 270–281, 2011.
    doi: 10.1016/b978-1-4160-6069-7.00026-2google scholar: lookup
  9. Makita S, Murakami N, Sakaguchi M, Furusho J. Development of horse-type quadruped robot. vol. 6, 930–935, 1999.
    doi: 10.1109/icsmc.1999.816677google scholar: lookup
  10. Moro FL. Horse-like walking, trotting, and galloping derived from kinematic motion primitives (kmps) and their application to walk/trot transitions in a compliant quadruped robot. Biological cybernetics 2013;107:309–320.
    doi: 10.1007/s00422-013-0551-9pubmed: 23463501google scholar: lookup
  11. Hoyt DF, Taylor CR. Gait and the energetics of locomotion in horses. Nature 1981;292:239–240.
    doi: 10.1038/292239a0google scholar: lookup
  12. Park HO, Dibazar AA, Berger TW. Cadence analysis of temporal gait patterns for seismic discrimination between human and quadruped footsteps. 1749–1752, 2009.
    doi: 10.1109/icassp.2009.4959942google scholar: lookup
  13. Buchner H, Obermüller S, Scheidl M. Body centre of mass movement in the lame horse. Equine Veterinary Journal 2001;33:122–127.
    doi: 10.1111/j.2042-3306.2001.tb05374.xpubmed: 11721552google scholar: lookup
  14. Rhodin M. Vertical movement symmetry of the withers in horses with induced forelimb and hindlimb lameness at trot. Equine veterinary journal 2018;50:818–824.
    doi: 10.1111/evj.12844pmc: PMC6175082pubmed: 29658147google scholar: lookup
  15. Ionescu C, Papava D, Olaru V, Sminchisescu C. Human3.6 m: Large scale datasets and predictive methods for 3d human sensing in natural environments. IEEE Transactions on Pattern Analysis and Machine Intelligence 2013;36:1325–1339.
    doi: 10.1109/TPAMI.2013.248pubmed: 26353306google scholar: lookup
  16. Sigal L, Balan A, Black M. Humaneva: Synchronized video and motion capture dataset and baseline algorithm for evaluation of articulated human motion. International Journal of Computer Vision 2010;87:4–27.
    doi: 10.1007/s11263-009-0273-6google scholar: lookup
  17. Mandery C, Terlemez Ö, Do M, Vahrenkamp N, Asfour T. The kit whole-body human motion database. 329–336, 2015.
    doi: 10.1109/icar.2015.7251476google scholar: lookup
  18. Santos G, Wanderley M, Tavares T, Rocha A. A multi-sensor human gait dataset captured through an optical system and inertial measurement units. Scientific Data 2022;9:545.
    doi: 10.1038/s41597-022-01638-2pmc: PMC9452504pubmed: 36071060google scholar: lookup
  19. Cao Z, Simon T, Wei S-E, Sheikh Y. Realtime multi-person 2d pose estimation using part affinity fields. 7291–7299, 2017.
    doi: 10.1109/cvpr.2017.143google scholar: lookup
  20. Mathis A. Deeplabcut: markerless pose estimation of user-defined body parts with deep learning. 2018.
    pubmed: 30127430doi: 10.1038/s41593-018-0209-ygoogle scholar: lookup
  21. Cao J. Cross-domain adaptation for animal pose estimation. 9498–9507, 2019.
    doi: 10.1109/iccv.2019.00959google scholar: lookup
  22. Kocabas M, Karagoz S, Akbas E. Self-supervised learning of 3d human pose using multi-view geometry. 1077–1086, 2019.
    doi: 10.1109/cvpr.2019.00117google scholar: lookup
  23. Li X, Fan Z, Liu Y, Li Y, Dai Q. 3d pose detection of closely interactive humans using multi-view cameras. Sensors 2019;19:2831.
    doi: 10.3390/s19122831pmc: PMC6631029pubmed: 31242651google scholar: lookup
  24. Joska D. Acinoset: a 3d pose estimation dataset and baseline models for cheetahs in the wild. 13901–13908, 2021.
    doi: 10.1109/icra48506.2021.9561338google scholar: lookup
  25. Günel S. Deepfly3d, a deep learning-based approach for 3d limb and appendage tracking in tethered, adult drosophila. Elife 2019;8:e48571.
    doi: 10.7554/eLife.48571pmc: PMC6828327pubmed: 31584428google scholar: lookup
  26. Bala PC. Automated markerless pose estimation in freely moving macaques with openmonkeystudio. Nature communications 2020;11:4560.
    doi: 10.1038/s41467-020-18441-5pmc: PMC7486906pubmed: 32917899google scholar: lookup
  27. Patel M, Gu Y, Carstensen LC, Hasselmo ME, Betke M. Animal pose tracking: 3d multimodal dataset and token-based pose optimization. International Journal of Computer Vision 2023;131:514–530.
    doi: 10.1007/s11263-022-01714-5google scholar: lookup
  28. Zimmermann C, Schneider A, Alyahyay M, Brox T, Diester I. Freipose: a deep learning framework for precise animal motion capture in 3d spaces. 2020–02.
    doi: 10.1101/2020.02.27.967620google scholar: lookup
  29. Gosztolai A. Liftpose3d, a deep learning-based approach for transforming two-dimensional to three-dimensional poses in laboratory animals. Nature methods 2021;18:975–981.
    doi: 10.1038/s41592-021-01226-zpmc: PMC7611544pubmed: 34354294google scholar: lookup
  30. Loper M, Mahmood N, Romero J, Pons-Moll G, Black MJ. Smpl: A skinned multi-person linear model. ACM Transactions on Graphics 2015;34:1–16.
    doi: 10.1145/2816795.2818013google scholar: lookup
  31. Mahmood N, Ghorbani N, Troje NF, Pons-Moll G, Black MJ. Amass: Archive of motion capture as surface shapes. 5442–5451, 2019.
    doi: 10.1109/iccv.2019.00554google scholar: lookup
  32. Kocabas M, Athanasiou N, Black MJ. Vibe: Video inference for human body pose and shape estimation. 5253–5263, 2020.
    doi: 10.1109/cvpr42600.2020.00530google scholar: lookup
  33. Ghorbani N, Black MJ. Soma: Solving optical marker-based mocap automatically. 11097–11106, 2021.
    doi: 10.1109/iccv48922.2021.01093google scholar: lookup
  34. Yuan Y, Iqbal U, Molchanov P, Kitani K, Kautz J. Glamr: Global occlusion-aware human mesh recovery with dynamic cameras. 11028–11039, 2022.
    doi: 10.1109/cvpr52688.2022.01076google scholar: lookup
  35. Li J. Task-generic hierarchical human motion prior using vaes. 771–781, 2021.
    doi: 10.1109/3dv53792.2021.00086google scholar: lookup
  36. Rempe D. Humor: 3d human motion model for robust pose estimation. 2021.
    doi: 10.1109/iccv48922.2021.01129google scholar: lookup
  37. Voleti V. Smpl-ik: Learned morphology-aware inverse kinematics for ai driven artistic workflows. SA ‘22, 2022.
    doi: 10.1145/3550340.3564227google scholar: lookup
  38. Zuffi S, Kanazawa A, Jacobs DW, Black MJ. 3d menagerie: Modeling the 3d shape and pose of animals. 6365–6373, 2017.
    doi: 10.1109/cvpr.2017.586google scholar: lookup
  39. Biggs B, Boyne O, Charles J, Fitzgibbon A, Cipolla R. Who left the dogs out? 3D animal reconstruction with expectation maximization in the loop. 2020.
    doi: 10.1007/978-3-030-58621-8_12google scholar: lookup
  40. Rüegg N, Zuffi S, Schindler K, Black MJ. Barc: Learning to regress 3d dog shape from images by exploiting breed information. 3876–3884, 2022.
    doi: 10.1109/cvpr52688.2022.00385google scholar: lookup
  41. Li C. hsmal: Detailed horse shape and pose reconstruction for motion pattern recognition. 2021.
    doi: 10.48550/arxiv.2106.10102google scholar: lookup
  42. Luo H. Artemis: articulated neural pets with appearance and motion synthesis. ACM Transactions on Graphics 2022;41:1–19.
    doi: 10.1145/3528223.3530086google scholar: lookup
  43. Dunn TW. Geometric deep learning enables 3d kinematic profiling across species and environments. Nature methods 2021;18:564–573.
    doi: 10.1038/s41592-021-01106-6pmc: PMC8530226pubmed: 33875887google scholar: lookup
  44. Zhang H, Starke S, Komura T, Saito J. Mode-adaptive neural networks for quadruped motion control. ACM Transactions on Graphics 2018;37:1–11.
    doi: 10.1145/3197517.3201366google scholar: lookup
  45. Kearney S, Li W, Parsons M, Kim KI, Cosker DR. Rgbd-dog: Predicting canine pose from rgbd sensors. 8336–8345, 2020.
    doi: 10.1109/cvpr42600.2020.00836google scholar: lookup
  46. Warner S, Koch T, Pfau T. Inertial sensors for assessment of back movement in horses during locomotion over ground. Equine Veterinary Journal 2010;42:417–424.
    doi: 10.1111/j.2042-3306.2010.00200.xpubmed: 21059039google scholar: lookup
  47. Bragança FS, Rhodin M, Van Weeren P. On the brink of daily clinical application of objective gait analysis: What evidence do we have so far from studies using an induced lameness model. The Veterinary Journal 2018;234:11–23.
    doi: 10.1016/j.tvjl.2018.01.006pubmed: 29680381google scholar: lookup
  48. Unt V, Evans J, Reed S, Pfau T, Weller R. Variation in frontal plane joint angles in horses. Equine Veterinary Journal 2010;42:444–450.
    doi: 10.1111/j.2042-3306.2010.00192.xpubmed: 21059043google scholar: lookup
  49. Bosch S. Equimoves: A wireless networked inertial measurement system for objective examination of horse gait. Sensors 2018;18:850.
    doi: 10.3390/s18030850pmc: PMC5877382pubmed: 29534022google scholar: lookup
  50. Ericson C, Stenfeldt P, Hardeman A, Jacobson I. The effect of kinesiotape on flexion-extension of the thoracolumbar back in horses at trot. Animals 2020;10:301.
    doi: 10.3390/ani10020301pmc: PMC7071056pubmed: 32069962google scholar: lookup
  51. Li V. The Poses for Equine Research Dataset (PFERD). Harvard Dataverse 2024.
    doi: 10.7910/DVN/2EXONEgoogle scholar: lookup
  52. Loper M, Mahmood N, Black MJ. Mosh: motion and shape capture from sparse markers. ACM Transactions on Graphics 2014;33:220–1.
    doi: 10.1145/2661229.2661273google scholar: lookup
  53. Yang J. Track anything: Segment anything meets videos. 2023.
    doi: 10.48550/arxiv.2304.11968google scholar: lookup
  54. Kirillov A. Segment anything. 2023.
    doi: 10.1109/iccv51070.2023.00371google scholar: lookup
  55. Sisson, S. (Saunders, 1911).

Citations

This article has been cited 1 times.
  1. Shaffer SK, Medjaouri O, Swenson B, Eliason T, Nicolella DP. A Markerless Approach for Full-Body Biomechanics of Horses.. Animals (Basel) 2025 Aug 5;15(15).
    doi: 10.3390/ani15152281pubmed: 40805071google scholar: lookup
Subscribe to our newsletter
Join 99,000 horse owners like you who receive our email updates on horse health and nutrition.