A Hierarchical Static-Dynamic Encoder-Decoder Structure for 3D Human Motion Prediction with Residual CNNs

Table 2

Influence of the v-CMU units on both datasets.

Model	G3D		FNTU
	MSE	MAE	MSE	MAE
PISEP^2 [20]	0.1199	1.1101	0.1210	1.1651
SDnet	0.1106	0.9782	0.1131	1.0675
Remove v-CMU	0.1236	1.0216	0.1209	1.1143