Presentation Information
[B-15-34]Detection of Poor Posture in Multiple People Using 3D Point Clouds of their Upper Body
◎Keitatsu Suzuki1, Hiroaki Morino1 (1. Graduate school of Engineering and Science, Shibaura Institute of Technology)
Keywords:
3D point clouds,Estimation of seated posture,LiDAR
It has been pointed out that maintaining poor posture—such as slouching or sitting with the sacrum on the floor—for long periods while performing desk work can have adverse health effects, and there is a need for technology that can easily detect such postures. Conventional methods, such as pressure sensors installed in seat cushions or image data captured by cameras, have been proposed [1]; however, these involve the hassle of installing and maintaining multiple sensors on each chair, as well as privacy concerns associated with prolonged recording. Therefore, we focus on analysis using 3D point cloud data acquired by a range-finding sensor called LiDAR. Since LiDAR does not contain color information, it does not include personally identifiable information, enabling privacy-conscious operation. Furthermore, by leveraging its ability to capture a person’s silhouette as a 3D shape, it can directly detect postural changes in the depth direction compared to methods based on 2D image data. This allows a single device to measure multiple people simultaneously, which is expected to reduce installation costs.
In this study, we propose a method that uses 3D point cloud data acquired by LiDAR to detect poor posture in multiple people based on upper-body shape features, even in situations where chairs and human bodies are captured as a single point cloud.
Translated with DeepL.com (free version)
In this study, we propose a method that uses 3D point cloud data acquired by LiDAR to detect poor posture in multiple people based on upper-body shape features, even in situations where chairs and human bodies are captured as a single point cloud.
Translated with DeepL.com (free version)