Skeleton Estimation in Gymnastics Using Segmentation Masks
Naoki Teramoto, Youji Ochi · 2024
Recently, data analysis using AI (Artificial Intelligence) and image recognition have attracted much attention. This study focuses on skeletal estimation in gymnastics. Since the skeleton of gymnasts is often mis-estimated, we decided to supplement the skeleton estimation by using segmentation masks. The accuracy of the skeletal estimation in this study was compared with the provisional true data. The accuracy was generally improved, but there were some problems, such as incorrect overlapping of some skeletons. We plan to improve the system and apply it to an automatic scoring system for gymnastics.