Skip to main content

Table 9 Results of the comparison with the dataset synthesized by domain-specific background images

From: First-person reading activity recognition by deep learning with synthetically generated images

 

Synthesized background images

 
 

Domain-specific

ImageNet

 
 

Others

Reading

Others

Reading

 

Scores

Precision

68.0

92.2

70.8

92.9

for each class

Recall

57.3

95.3

69.9

95.0

 

F-measure

57.1

93.5

64.9

93.5

Averages

Precision

81.2

81.9

over classes

Recall

72.2

82.5

 

F-measure

73.1

79.2

  1. Best methods for a given measure are specified in italic type