Visual Computing for Industry, Biomedicine, and Art

Table 3 The ablation experiments for different module parameters in STTG-Net, with reported results for joint angle errors on Human3.6M

From: STTG-net: a Spatio-temporal network for human motion prediction based on transformer and graph convolution network

H	8	6	12	8	8	8	8
L_T	6	6	6	4	8	6	6
L_S	14	14	14	14	14	12	16
80 ms	0.27	0.28	0.28	0.28	0.28	0.29	0.27
160 ms	0.52	0.55	0.53	0.53	0.55	0.54	0.53
320 ms	0.86	0.87	0.86	0.86	0.89	0.87	0.87
400 ms	0.98	0.99	0.99	0.98	1.01	1.00	0.99

The best results are presented in bold

Back to article page