Nonnegative Tensor-Based Linear Dynamical Systems for Recognizing Human Action from 3D Skeletons

Table 8

Comparison between LTBSVM and infinite Grassmann manifold.

Dataset	Grassmann manifold[8] (ARMA Model + LTBSVM)	ARMA model + Infinite Grassmann Manifold
MSR-Action3D	91.21	94.51
UTKinect-Action	88.5	90.65
G3D-Gaming	85.26	89.13