@article{ART003204600},
author={Jae-Seung Kim and Hyung-Woo Moon and Yong-Tae Woo},
title={Outcome-Based Comparison of Representative Pitching Motions Using Deep Learning–Based 3D Skeleton Sequences},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2025},
volume={30},
number={5},
pages={79-87}
TY - JOUR
AU - Jae-Seung Kim
AU - Hyung-Woo Moon
AU - Yong-Tae Woo
TI - Outcome-Based Comparison of Representative Pitching Motions Using Deep Learning–Based 3D Skeleton Sequences
JO - Journal of The Korea Society of Computer and Information
PY - 2025
VL - 30
IS - 5
PB - The Korean Society Of Computer And Information
SP - 79
EP - 87
SN - 1598-849X
AB - In this paper, we propose a method to define representative pitching motions centered on pitching results in baseball games and proposes a method to analyze the influence of pitch type or pitching results on pitching motions. To this end, the videos taken from the front of the pitcher were classified into four types, and the pitching videos were analyzed by type. The pitching videos were divided into frames, and a 2D skeleton was extracted using HRNet, and then converted into a time-series 3D skeleton using PoseFormerV2. The relative coordinates based on the pelvis were normalized to correct the shooting distortion. DTW Barycenter Averaging (DBA) was applied to generate representative pitching motions by type, and the similarity between representative pitching motions was compared using the Dynamic Time Warping technique. The experimental results showed that the difference in pitching motions according to the pitching results was more significant, which means that individual pitchers have result-centered pitching motions.
KW - Deep Learning;HRNet;PoseFormerV2;Pitching motion analysis;DTW
DO -
UR -
ER -
Jae-Seung Kim, Hyung-Woo Moon and Yong-Tae Woo. (2025). Outcome-Based Comparison of Representative Pitching Motions Using Deep Learning–Based 3D Skeleton Sequences. Journal of The Korea Society of Computer and Information, 30(5), 79-87.
Jae-Seung Kim, Hyung-Woo Moon and Yong-Tae Woo. 2025, "Outcome-Based Comparison of Representative Pitching Motions Using Deep Learning–Based 3D Skeleton Sequences", Journal of The Korea Society of Computer and Information, vol.30, no.5 pp.79-87.
Jae-Seung Kim, Hyung-Woo Moon, Yong-Tae Woo "Outcome-Based Comparison of Representative Pitching Motions Using Deep Learning–Based 3D Skeleton Sequences" Journal of The Korea Society of Computer and Information 30.5 pp.79-87 (2025) : 79.
Jae-Seung Kim, Hyung-Woo Moon, Yong-Tae Woo. Outcome-Based Comparison of Representative Pitching Motions Using Deep Learning–Based 3D Skeleton Sequences. 2025; 30(5), 79-87.
Jae-Seung Kim, Hyung-Woo Moon and Yong-Tae Woo. "Outcome-Based Comparison of Representative Pitching Motions Using Deep Learning–Based 3D Skeleton Sequences" Journal of The Korea Society of Computer and Information 30, no.5 (2025) : 79-87.
Jae-Seung Kim; Hyung-Woo Moon; Yong-Tae Woo. Outcome-Based Comparison of Representative Pitching Motions Using Deep Learning–Based 3D Skeleton Sequences. Journal of The Korea Society of Computer and Information, 30(5), 79-87.
Jae-Seung Kim; Hyung-Woo Moon; Yong-Tae Woo. Outcome-Based Comparison of Representative Pitching Motions Using Deep Learning–Based 3D Skeleton Sequences. Journal of The Korea Society of Computer and Information. 2025; 30(5) 79-87.
Jae-Seung Kim, Hyung-Woo Moon, Yong-Tae Woo. Outcome-Based Comparison of Representative Pitching Motions Using Deep Learning–Based 3D Skeleton Sequences. 2025; 30(5), 79-87.
Jae-Seung Kim, Hyung-Woo Moon and Yong-Tae Woo. "Outcome-Based Comparison of Representative Pitching Motions Using Deep Learning–Based 3D Skeleton Sequences" Journal of The Korea Society of Computer and Information 30, no.5 (2025) : 79-87.
