Table 2 Comparison of detection accuracy between FACSNet and SRNet (JPEG domain) %.
From: FACSNet: Forensics aided content selection network for heterogeneous image steganalysis
Dataset | Steganography algorithm | Detection model | QF75 | QF95 | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
0.1 bpnzAC | 0.2 bpnzAC | 0.3 bpnzAC | 0.4 bpnzAC | 0.1 bpnzAC | 0.2 bpnzAC | 0.3 bpnzAC | 0.4 bpnzAC | |||
BOSSbase (All Original) | J-UNIWARD | SRNet | 64.49 | 79.28 | 86.67 | 91.75 | 59.66 | 68.83 | 75.44 | 80.46 |
FACSNet | 64.49 | 79.31 | 86.67 | 91.75 | 59.66 | 68.83 | 75.44 | 80.46 | ||
UERD | SRNet | 81.61 | 89.85 | 94.33 | 96.58 | 67.60 | 78.41 | 85.73 | 89.38 | |
FACSNet | 81.61 | 89.85 | 94.34 | 96.58 | 67.63 | 78.41 | 85.73 | 89.38 | ||
Heterogeneous-BOSSbase (Original + AI) | J-UNIWARD | SRNet | 60.22 | 75.34 | 81.21 | 85.30 | 51.73 | 62.91 | 70.04 | 77.60 |
FACSNet | 64.38 | 78.85 | 86.63 | 91.73 | 58.35 | 68.56 | 75.09 | 80.37 | ||
UERD | SRNet | 75.83 | 86.66 | 89.01 | 93.18 | 61.25 | 72.91 | 81.46 | 86.27 | |
FACSNet | 81.58 | 89.70 | 93.66 | 96.50 | 67.55 | 78.33 | 85.71 | 89.38 | ||
BOWS (All Original) | J-UNIWARD | SRNet | 64.13 | 77.08 | 85.77 | 90.34 | 56.84 | 67.69 | 74.28 | 79.88 |
FACSNet | 64.13 | 77.09 | 85.77 | 90.34 | 56.85 | 67.69 | 74.32 | 79.88 | ||
UERD | SRNet | 80.77 | 89.48 | 93.60 | 95.43 | 67.55 | 78.03 | 85.34 | 88.69 | |
FACSNet | 80.79 | 89.48 | 93.60 | 95.43 | 67.55 | 78.03 | 85.34 | 88.69 | ||
Heterogeneous-BOWS (Original + AI) | J-UNIWARD | SRNet | 59.81 | 72.55 | 80.97 | 85.63 | 50.11 | 61.75 | 70.38 | 73.66 |
FACSNet | 63.39 | 76.80 | 85.52 | 90.18 | 56.34 | 67.65 | 74.19 | 79.83 | ||
UERD | SRNet | 74.07 | 84.81 | 88.62 | 91.83 | 62.33 | 70.80 | 80.77 | 85.04 | |
FACSNet | 80.66 | 89.20 | 93.44 | 95.41 | 67.39 | 77.92 | 85.18 | 88.65 |
