notesum.ai

Published at November 29

SkelMamba: A State Space Model for Efficient Skeleton Action Recognition of Neurological Disorders

cs.CV

cs.AI

cs.LG

Released Date: November 29, 2024

Authors: Niki Martinel¹, Mariano Serrao², Christian Micheloni¹

Aff.: ¹University of Udine; ²Sapienza Universita di Roma

Arxiv: http://arxiv.org/pdf/2411.19544v1

Refer to caption

		ND			KOA-PD-NM			KOA-PD-NM-Severity
Type	Model	$\mathbb{M}_{j}$	$\mathbb{M}_{jb}$	$\mathbb{M}_{jbm}$	$\mathbb{M}_{j}$	$\mathbb{M}_{jb}$	$\mathbb{M}_{jbm}$	$\mathbb{M}_{j}$	$\mathbb{M}_{jb}$	$\mathbb{M}_{jbm}$
CNN	PoseC3D [13]	98.95	99.21	99.35	84.79	89.40	89.86	95.79	98.95	98.95
GCN	2S-AAGCN [48]	94.74	97.89	98.95	91.22	92.39	93.18	89.47	93.68	93.68
	STGCN++ [65]	93.68	94.74	95.79	93.54	94.31	95.32	95.32	95.79	96.84
Transformer	Hyperformer [75]	99.21	99.35	99.21	96.82	96.94	98.14	95.79	96.84	98.95
SSM	SkelMamba (Ours)	99.35	99.35	99.64	97.45	98.23	98.62	96.84	98.95	99.21