Table 5 Confusion matrix for the MSR-Action3D dataset (SS2)
From: Flexible human action recognition in depth video sequences using masked joint trajectories
(a) | (d) | (g) | (h) | (i) | (k) | (l) | (n) | |
|---|---|---|---|---|---|---|---|---|
(a) | 83.33 % (10/12) | 8.33 % (1/12) | 8.33 % (1/12) | |||||
(d) | 50 % (6/12) | 16.67 % (2/12) | 16.67 % (2/12) | 16.67 % (2/12) | ||||
(g) | 92.31 % (12/13) | 7.69 % (1/13) | ||||||
(h) | 20 % (3/15) | 80 % (12/15) | ||||||
(i) | 26.67 % (4/15) | 13.33 % (2/15) | 60 % (9/15) | |||||
(k) | 100 % (15/15) | |||||||
(l) | 6.66 % (1/15) | 86.68 % (13/15) | 6.66 % (1/15) | |||||
(n) | 100 % (15/15) |