EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

MuTr: Multi-Stage Transformer for Hand Pose Estimation from Full-Scene Depth Image

Jakub Kanis iD, Ivan Gruber iD, Zdeněk Krňoul iD, Matyáš Boháček iD, Jakub Straka iD, Marek Hrúz iD

Sensors · 12.06.2023 · 23 (12) · 5509

DOI: 10.3390/s23125509 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 12.06.2023. Verfasst von Jakub Kanis und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Cybernetics and New Technologies for the Information Society, University of West Bohemia Technická 8, 301 00 Pilsen, Czech Republic. Hauptthemen laut MEDLINE: Hand, Upper Extremity. Die Arbeit wurde bisher 8-mal zitiert (Crossref) und nennt 50 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: University of West Bohemia, European Regional Development Fund, Regional Development Fund.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This work presents a novel transformer-based method for hand pose estimation—DePOTR. We test the DePOTR method on four benchmark datasets, where DePOTR outperforms other transformer-based methods while achieving results on par with other state-of-the-art methods. To further demonstrate the strength of DePOTR, we propose a novel multi-stage approach from full-scene depth image—MuTr. MuTr removes the necessity of having two different models in the hand pose estimation pipeline—one for hand localization and one for pose estimation—while maintaining promising results. To the best of our knowledge, this is the first successful attempt to use the same model architecture in standard and simultaneously in full-scene image setup while achieving competitive results in both of them. On the NYU dataset, DePOTR and MuTr reach precision equal to 7.85 mm and 8.71 mm, respectively.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

HandUpper ExtremityBenchmarkingElectric Power SuppliesKnowledge

Stichwörter der Autor:innen

hand pose estimationmulti-stageneural networktransformer

Autor:innen und Institutionen

  • Department of Cybernetics and New Technologies for the Information Society, University of West Bohemia Technická 8, 301 00 Pilsen, Czech Republic.
  • Gymnasium of Johannes Kepler, Parléřova 2/118, 169 00 Prague, Czech Republic.

ORCID-Profile: Jakub Kanis, Ivan Gruber, Zdeněk Krňoul, Matyáš Boháček, Jakub Straka, Marek Hrúz

Förderung

  • University of West Bohemia
  • European Regional Development Fund
  • Regional Development Fund
  • National Grid Infrastructure MetaCentrum

Zitieren

APA

Kanis, J., Gruber, I., Krňoul, Z., Boháček, M., Straka, J., & Hrúz, M. (2023). MuTr: Multi-Stage Transformer for Hand Pose Estimation from Full-Scene Depth Image. Sensors, 23(12), 5509. https://doi.org/10.3390/s23125509

Vancouver

Kanis J, Gruber I, Krňoul Z, Boháček M, Straka J, Hrúz M. MuTr: Multi-Stage Transformer for Hand Pose Estimation from Full-Scene Depth Image. Sensors. 2023;23(12):5509. doi:10.3390/s23125509

BibTeX
@article{Kanis2023,
  author = {Kanis, Jakub and Gruber, Ivan and Krňoul, Zdeněk and Boháček, Matyáš and Straka, Jakub and Hrúz, Marek},
  title = {MuTr: Multi-Stage Transformer for Hand Pose Estimation from Full-Scene Depth Image},
  journal = {Sensors},
  year = {2023},
  volume = {23},
  number = {12},
  pages = {5509},
  doi = {10.3390/s23125509}
}
Kostenlos starten →