notesum.ai

Published at November 25

SMGDiff: Soccer Motion Generation using diffusion probabilistic models

cs.CV

Released Date: November 25, 2024

Authors: Hongdi Yang¹, Chengyang Li¹, Zhenxuan Wu¹, Gaozheng Li¹, Jingya Wang¹, Jingyi Yu¹, Zhuo Su², Lan Xu¹

Aff.: ¹ShanghaiTech University; ²ByteDance

Arxiv: http://arxiv.org/abs/2411.16216v1

Refer to caption

	Motion Quality				Condition Alignment
Method	FID $\downarrow$	Ft. Slid. $\downarrow$	Accel. $\downarrow$	Div. $\uparrow$	Traj. Err. $\downarrow$	Orient. Err. $\downarrow$	Skill Acc. $\uparrow$
LMP [57]	0.3541	1.0678	1.6070	0.3980	4.1156	6.4932	73.3%
MANN-DP [72, 59]	0.3593	1.3507	1.5652	0.4753	4.0690	5.2985	69.1%
CM [60]	0.2494	1.6498	1.1752	0.3520	3.1034	5.0663	52.9%
Ours	0.1813	0.8543	1.1999	0.6177	2.4132	4.9393	93.3%