Table 2 Comparison of proposed method with the state of the art on KTH dataset
From: Spatio-temporal information for human action recognition
Methods | KTH |
|---|---|
Fathi et al. 2008 [33] | 90.50% |
Laptev et al. 2008 [2] | 91.80% |
Iosifidis et al. 2014 [34] | 92.13% |
Zhen and Shao 2013 [8] | 92.20% |
Bregonzio et al. 2009 [35] | 93.17% |
Yuan et al., 2009 [36] | 93.30% |
Liu et al. 2009 [4] | 93.80% |
Zhou et al., 2014 [9] | 93.80% |
Wang et al. 2011 [3] | 93.80% |
Wang et al, 2013 [29] | 94.20% |
Gilbert et al. 2009 [17] | 94.50% |
Kovashka and Grauman, 2010 [7] | 94.53% |
MBH + POOL STRep | 95.37% |
MBH + CP STRep | 94.91% |
STBi-fusion + BoW | 94.44% |
STBi-fusion + CP STRep | 95.37% |
STBi-fusion + POOL STRep | 95.83% |