Skip to main content
Cybersecurity

Table 8 Detection accuracy comparison with and without some components to measure the impact of each component on the system performance

From: A convolutional neural network to detect possible hidden data in spatial domain images

Architecture

Detection accuracy

WOW

S-UNIWARD

0.2 bpp

0.4 bpp

0.2 bpp

0.4 bpp

Proposed method

90.2

94.4

79.3

93.1

Proposed without 2D depthwise

81.6

90.2

74.9

88.2

Proposed without LReLu

83.4

91.7

77.4

90.1

Proposed without Multi-scale pooling

79.2

87.9

74.1

86.2

Proposed without 2D depthwise and LReLu

79.8

88.3

75.5

87.4

Proposed without 2D depthwise and multi-scale pooling

77.1

85.6

73.2

84.7

Proposed without LReLu and Multi-scale pooling

78.2

87.0

72.4

86.3

Back to article page