Skip to main content

Table 9 Comparison of different methods on their respective datasets

From: A multisource fusion framework driven by user-defined knowledge for egocentric activity recognition

 

Proposed method (%)

Method proposed in [15] (ConvNets+LSTM) + pooling fusion (%)

Method proposed in [16] (DT + temporal enhanced features) + Fisher vector (%)

Average pooling

Maximum pooling

Vision

79.2

68.5%

75.0

78.4

Sensors

43.1

49.5

69.0

Fusion

85.4

76.5

80.5

83.7

  1. Proposed method was applied to the eButton datasets described in Section 3.1; the other two methods were applied to the datasets described in [16]