Table 3 Comparison of proposed method with the state of the art on Olympic dataset

Laptev et al, 2008 [2]

62.50%

Tang et al. 2012 [27]

66.80%

Niebles et al. 2010 [28]

72.10%

Wang et al. 2013 [29]

71.60%

Liu et al. 2011 [30]

74.38%

STBi-fusion + BoW

71.48%

STBi-fusion + POOL STRep

73.21%

STBi-fusion + CP STRep

74.40%