Pose Estimation Analysis and Fine-Tuning on the REHAB24-6
Rehabilitation Dataset

J 2025

Pose Estimation Analysis and Fine-Tuning on the REHAB24-6 Rehabilitation Dataset

ČERNEK, Andrej; Jan SEDMIDUBSKÝ a Petra BUDÍKOVÁ

Základní údaje

Originální název

Pose Estimation Analysis and Fine-Tuning on the REHAB24-6 Rehabilitation Dataset

Autoři

ČERNEK, Andrej; Jan SEDMIDUBSKÝ a Petra BUDÍKOVÁ

Vydání

Information Systems, 2025, 0306-4379

Další údaje

Jazyk

angličtina

Typ výsledku

Článek v odborném periodiku

Obor

10200 1.2 Computer and information sciences

Stát vydavatele

Dánsko

Utajení

není předmětem státního či obchodního tajemství

Odkazy

URL

Impakt faktor

Impact factor: 3.400 v roce 2024

Organizační jednotka

Fakulta informatiky

DOI

https://doi.org/10.1016/j.is.2025.102579

UT WoS

001537388900001

Klíčová slova anglicky

REHAB24-6 dataset; pose estimation; motion capture; rehabilitation exercise; skeleton format; fine-tuning 2D/3D detectors; similarity of repetitions

Štítky

DISA

Příznaky

Mezinárodní význam, Recenzováno

Změněno: 29. 1. 2026 21:18, doc. RNDr. Jan Sedmidubský, Ph.D.

Anotace

V originále

Human motion analysis is a key enabler for remote healthcare applications, particularly in physical rehabilitation. In this context, mobile devices equipped with RGB cameras seem to be a promising technology for monitoring patients during home-based exercises and providing real-time feedback. This relies on pose estimation algorithms that extract spatio-temporal features of human motion from video data. While state-of-the-art models can estimate body pose from mobile video streams, their effectiveness in rehabilitation scenarios remains underexplored. To address this, we introduce the REHAB24-6 dataset, which includes untrimmed RGB videos, 2D and 3D skeletal ground truth annotations, and temporal segmentation for six common rehabilitation exercises. We also propose an evaluation protocol for assessing different aspects of quality of pose estimation methods, dealing with challenges that arise when different skeleton formats are compared. Additionally, we show how fine-tuning of existing models on our dataset leads to improved quality. Our experimental results compare several state-of-the-art approaches and highlight their key limitations -- particularly in depth estimation -- offering practical insights for selecting and improving pose estimation systems for rehabilitation monitoring.

Návaznosti

FW09020055, projekt VaV

Název: VisioTherapy: Podpora fyzioterapeutické léčby pomocí počítačové analýzy pohybu

Investor: Technologická agentura ČR, VisioTherapy: Podpora fyzioterapeutické léčby pomocí počítačové analýzy pohybu, Podprogram 2 "Nováčci"

MUNI/A/1638/2024, interní kód MU

Název: Umělá inteligence a správa komplexních rozsáhlých dat

Investor: Masarykova univerzita, Umělá inteligence a správa komplexních rozsáhlých dat

Citovat

ČERNEK, Andrej; Jan SEDMIDUBSKÝ a Petra BUDÍKOVÁ. Pose Estimation Analysis and Fine-Tuning on the REHAB24-6 Rehabilitation Dataset. Information Systems. 2025, roč. 134, October–November, s. 1-12. ISSN 0306-4379. Dostupné z: https://doi.org/10.1016/j.is.2025.102579.

@article{2505537,
   author = {Černek, Andrej and Sedmidubský, Jan and Budíková, Petra},
   article_number = {October–November},
   doi = {https://doi.org/10.1016/j.is.2025.102579},
   keywords = {REHAB24-6 dataset; pose estimation; motion capture; rehabilitation exercise; skeleton format; fine-tuning 2D/3D detectors; similarity of repetitions},
   language = {eng},
   issn = {0306-4379},
   journal = {Information Systems},
   title = {Pose Estimation Analysis and Fine-Tuning on the REHAB24-6 Rehabilitation Dataset},
   url = {https://www.sciencedirect.com/science/article/pii/S0306437925000638},
   volume = {134},
   year = {2025}
}

TY  - JOUR
ID  - 2505537
AU  - Černek, Andrej - Sedmidubský, Jan - Budíková, Petra
PY  - 2025
TI  - Pose Estimation Analysis and Fine-Tuning on the REHAB24-6 Rehabilitation Dataset
JF  - Information Systems
VL  - 134
IS  - October–November
SP  - 1-12
EP  - 1-12
SN  - 03064379
KW  - REHAB24-6 dataset
KW  - pose estimation
KW  - motion capture
KW  - rehabilitation exercise
KW  - skeleton format
KW  - fine-tuning 2D/3D detectors
KW  - similarity of repetitions
UR  - https://www.sciencedirect.com/science/article/pii/S0306437925000638
N2  - Human motion analysis is a key enabler for remote healthcare applications, particularly in physical rehabilitation. In this context, mobile devices equipped with RGB cameras seem to be a promising technology for monitoring patients during home-based exercises and providing real-time feedback. This relies on pose estimation algorithms that extract spatio-temporal features of human motion from video data. While state-of-the-art models can estimate body pose from mobile video streams, their effectiveness in rehabilitation scenarios remains underexplored. To address this, we introduce the REHAB24-6 dataset, which includes untrimmed RGB videos, 2D and 3D skeletal ground truth annotations, and temporal segmentation for six common rehabilitation exercises. We also propose an evaluation protocol for assessing different aspects of quality of pose estimation methods, dealing with challenges that arise when different skeleton formats are compared. Additionally, we show how fine-tuning of existing models on our dataset leads to improved quality. Our experimental results compare several state-of-the-art approaches and highlight their key limitations -- particularly in depth estimation -- offering practical insights for selecting and improving pose estimation systems for rehabilitation monitoring.
ER  -

ČERNEK, Andrej; Jan SEDMIDUBSKÝ a Petra BUDÍKOVÁ. Pose Estimation Analysis and Fine-Tuning on the REHAB24-6 Rehabilitation Dataset. \textit{Information Systems}. 2025, roč.~134, October–November, s.~1-12. ISSN~0306-4379. Dostupné z: https://doi.org/10.1016/j.is.2025.102579.

Přehled o publikaci