Table 4 Confusion matrix for the MSR-Action3D dataset (SS1)
From: Flexible human action recognition in depth video sequences using masked joint trajectories
(b) | (c) | (e) | (f) | (j) | (m) | (r) | (t) | |
|---|---|---|---|---|---|---|---|---|
(b) | 50 % (6/12) | 8.33 % (1/12) | 41.67 % (5/12) | |||||
(c) | 75 % (9/12) | 25 % (3/12) | ||||||
(e) | 100 % (11/11) | |||||||
(f) | 18.18 % (2/11) | 9.09 % (1/11) | 72.73 % (8/11) | |||||
(j) | 100 % (15/15) | |||||||
(m) | 46.67 % (7/15) | 53.33 % (8/15) | ||||||
(r) | 100 % (15/15) | |||||||
(t) | 7.14 % (1/14) | 92.86 % (13/14) |