GS3-2 Seated Posture Estimation Based on Monocular Camera Images

Title:	GS3-2 Seated Posture Estimation Based on Monocular Camera Images
Publication:	ICAROB2025
Volume:	30
Pages:	845-849
ISSN:	2188-7829
DOI:	10.5954/ICAROB.2025.GS3-2
Author(s):	Hitoshi Shimomae, Tsubasa Esumi, Noriko Takemura
Publication Date:	February 13, 2025
Keywords:	Posture Analysis, CNN, GNN
Abstract:	Poor seated posture significantly strains the body, leading to symptoms like shoulder stiffness and back pain. While research on seated posture estimation using images has been active, many studies focus on extreme postures not typically seen in daily desk work. This study aims to estimate common postures, such as slouching, which are often experienced in everyday settings. Due to the lack of medical quantitative metrics for evaluating posture quality, we manually annotated some of our collected posture data and used semi-automatic annotation by SVM to build a dataset. Using this dataset, we trained deep learning models for posture estimation with different input data types: RGB images, silhouette images, and posture key points, and compared their performance.
PDF File:	https://alife-robotics.co.jp/members2025/icarob/data/html/data/GS/GS3/GS3-2.pdf
Copyright:	© The authors. This article is distributed under the terms of the Creative Commons Attribution License 4.0, which permits non-commercial use, distribution and reproduction in any medium, provided the original work is properly cited. See for details: https://creativecommons.org/licenses/by-nc/4.0/