Communication Dans Un Congrès Année : 2025

Impact of video resolution on center of mass estimation: A cross-models analysis (preliminary study)

David Pagnon

Résumé

Recent advances in markerless motion capture (MC) have made biomechanical analysis more accessible. However, video parameters might affect pose estimation accuracy. This study investigates how video quality (540p and 1080p at 16:9) influences 3D center of mass (CoM) accuracy across different triangulated 2D pose estimation algorithms (RTMpose, OpenPose, OpenPose with marker augmentation and Theia3D). A healthy participant (age: 52) performed 60 statics standings trials. CoM estimations from each algorithm were compared against Theia3D. Results indicate that video quality significantly affects estimation accuracy, with varying impacts across algorithms. Our findings suggest considering video settings in markerless MC.

Fichier principal
Vignette du fichier
ISB2025_VF.pdf (180.05 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05361226 , version 1 (12-11-2025)

Licence

Identifiants

  • HAL Id : hal-05361226 , version 1

Citer

Ahmed Yacine Lardjane, David Pagnon, Laura Wallard, Christophe Gillet, Christopher Paglia, et al.. Impact of video resolution on center of mass estimation: A cross-models analysis (preliminary study). Congress of the International Society of Biomechanics, International Society of Biomechanics, Jul 2025, Stockholm, Sweden. ⟨hal-05361226⟩
103 Consultations
38 Téléchargements

Partager

  • More