主催: 一般社団法人 日本機械学会
会議名: 2023年度 年次大会
開催日: 2023/09/03 - 2023/09/06
To digitally archive traditional Japanese dance, it is necessary to record the body movements of the performers wearing stage costumes. In conventional motion capture techniques, attaching special sensors to capture dance movements can be burdensome for performers. In recent years, markerless video motion analysis systems have been used for motion analysis. However, neither conventional nor markerless motion capture can accurately estimate body parts hidden by stage costumes. Therefore, in this work, we tried to utilize markerless motion capture to estimate motion while performers wear stage costumes by retraining existing pre-trained models.