Summary
International Workshop on Smart Info-Media Systems in Asia
2023
Session Number:RS3
Session:
Number:RS3-3
Human Pose Estimation Based on Part-Wise 6-DoF Pose Estimation
Yuki Nagoshi, Takeru Inoue, Junya Morioka, Ryusuke Miyamoto,
pp.77-81
Publication Date:2023/8/31
Online ISSN:2188-5079
DOI:10.34385/proc.77.RS3-3
PDF download
Summary:
Human pose estimation is a critical technology that can be applied to practical applications, such as improving sports performance, detecting abnormality in human behavior, and analyzing gait. Many existing methods estimate the twodimensional positions of joints in the image plane, whereas some also perform three-dimensional position estimation, but their accuracy still has room for improvement. Therefore, in this study, we propose a method for human pose estimation by applying a rigid-body six-degree-of-freedom estimation technique to individual parts of the human body. The experimental results using computer graphics demonstrated a 25% reduction in rotation errors compared to the existing methods.