EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Accuracy Evaluation of 3D Pose Reconstruction Algorithms Through Stereo Camera Information Fusion for Physical Exercises with MediaPipe Pose

Sebastian Dill iD, Arjang Ahmadi iD, Martin Grimmer iD, Dennis Haufe iD, Maurice Rohr iD, Yanhua Zhao iD, Maziar Sharbafi iD, Christoph Hoog Antink iD

Sensors · 04.12.2024 · 24 (23) · 7772

DOI: 10.3390/s24237772 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 04.12.2024. Verfasst von Sebastian Dill und 7 weiteren Autor:innen. Beteiligte Einrichtung: KIS*MED (AI Systems in Medicine), Technische Universität Darmstadt, 64283 Darmstadt, Germany. Hauptthemen laut MEDLINE: Algorithms, Imaging, Three-Dimensional, Exercise. Die Arbeit wurde bisher 42-mal zitiert (Crossref) und nennt 34 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Hessian Ministry for Digital Strategy and Development, TU Darmstadt Forum for Interdisciplinary Research, TU Darmstadt’s Forum for Interdisciplinary Research.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In recent years, significant research has been conducted on video-based human pose estimation (HPE). While monocular two-dimensional (2D) HPE has been shown to achieve high performance, monocular three-dimensional (3D) HPE poses a more challenging problem. However, since human motion happens in a 3D space, 3D HPE offers a more accurate representation of the human, granting increased usability for complex tasks like analysis of physical exercise. We propose a method based on MediaPipe Pose, 2D HPE on stereo cameras and a fusion algorithm without prior stereo calibration to reconstruct 3D poses, combining the advantages of high accuracy in 2D HPE with the increased usability of 3D coordinates. We evaluate this method on a self-recorded database focused on physical exercise to research what accuracy can be achieved and whether this accuracy is sufficient to recognize errors in exercise performance. We find that our method achieves significantly improved performance compared to monocular 3D HPE (median RMSE of 30.1 compared to 56.3, p-value below 10−6) and can show that the performance is sufficient for error recognition.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

AlgorithmsImaging, Three-DimensionalExerciseHumansPostureVideo Recording

Stichwörter der Autor:innen

MediaPipe Posecomputer visionhuman pose estimationinformation fusion

Autor:innen und Institutionen

  • KIS*MED (AI Systems in Medicine), Technische Universität Darmstadt, 64283 Darmstadt, Germany.
  • Lauflabor (Locomotion Laboratory), Centre for Cognitive Science, Technische Universität Darmstadt, 64289 Darmstadt, Germany.

ORCID-Profile: Sebastian Dill, Arjang Ahmadi, Martin Grimmer, Dennis Haufe, Maurice Rohr, Yanhua Zhao, Maziar Sharbafi, Christoph Hoog Antink

Förderung

  • Hessian Ministry for Digital Strategy and Development
  • TU Darmstadt Forum for Interdisciplinary Research
  • TU Darmstadt’s Forum for Interdisciplinary Research
  • Technische Universität Darmstadt and University
  • State Library Darmstadt

Zitieren

APA

Dill, S., Ahmadi, A., Grimmer, M., Haufe, D., Rohr, M., Zhao, Y., Sharbafi, M., & Hoog Antink, C. (2024). Accuracy Evaluation of 3D Pose Reconstruction Algorithms Through Stereo Camera Information Fusion for Physical Exercises with MediaPipe Pose. Sensors, 24(23), 7772. https://doi.org/10.3390/s24237772

Vancouver

Dill S, Ahmadi A, Grimmer M, Haufe D, Rohr M, Zhao Y, et al. Accuracy Evaluation of 3D Pose Reconstruction Algorithms Through Stereo Camera Information Fusion for Physical Exercises with MediaPipe Pose. Sensors. 2024;24(23):7772. doi:10.3390/s24237772

BibTeX
@article{Dill2024,
  author = {Dill, Sebastian and Ahmadi, Arjang and Grimmer, Martin and Haufe, Dennis and Rohr, Maurice and Zhao, Yanhua and Sharbafi, Maziar and Hoog Antink, Christoph},
  title = {Accuracy Evaluation of 3D Pose Reconstruction Algorithms Through Stereo Camera Information Fusion for Physical Exercises with MediaPipe Pose},
  journal = {Sensors},
  year = {2024},
  volume = {24},
  number = {23},
  pages = {7772},
  doi = {10.3390/s24237772}
}
Kostenlos starten →