Author(s): HSU, C.T., BAN, Y.R., CHEN, J.S., CHANG, Y.H, CHIU, T.Y., HO, W.H., Institution: UNIVERSITY OF TAIPEI, Country: CHINA, Abstract-ID: 1690

INTRODUCTION:
A novel female snatch performance index is proposed and verified in this paper. Many successful snatch measure indexes such as horizontal displacement and vertical displacement have been observed, however, these indexes need experts utilizing kinematic analysis software to gather. Hence, in this study, an open AI skeleton is considered to obtain the skeleton information automatically and further compose the 2D skeleton information to our proposed snatch performance index.
METHODS:
Intel OpenVino with Microsoft CoCo skeleton model is considered in this paper. We gather woman 55Kg and 59Kg categories snatch competition videos from 2019 IWF weightlifting world championship and 2021 Tokyo Olympic games. These videos are separated into elite and general groups. Skeleton model is utilized to obtain the lifter’s joint coordinate from the videos We refer to the snatch phases (Ikeda, et al, 2012; Musser, Garhammer, Rozenek, Crussemeyer, & Vargas, 2014) and further consider the joint passing time in each phase as our novel female snatch performance index. Finally, independent sample t test verifies our proposed index.
RESULTS:
From independent sample t test, the passing time of snatch phase III and phase IV are significant difference. Elite lifters consume less time in phase I and phase II than general lifters. Furthermore, the most-high level of the barbell of the elite lifters is lower than general lifters.
CONCLUSION:
In this paper, we proposed a novel female snatch performance index to indicate the female weightlifting lifters’ performance. From the results, we can obtain that our proposed index indicates that elite lifters consume less time in phase III and phase IV and utilize less most-high level of the barbell. These observation shows that our proposed female snatch performance index can efficiently and precisely indicate lifters’ performance.