Table 3 The performance of Linear Kernel-Net, Non-linear Kernel-Net, and All Kernel-Net on resampled images
From: Joint multi-domain feature learning for image steganalysis based on CNN
Algorithms | Linear Kernel-Net | Non-linear Kernel-Net | All Kernel-Net |
|---|---|---|---|
WOW (0.2 bpp) | 0.710 | 0.697 | 0.714 |
WOW (0.4 bpp) | 0.819 | 0.788 | 0.844 |
S-UNIWARD (0.2 bpp) | 0.661 | 0.609 | 0.669 |
S-UNIWARD (0.4 bpp) | 0.766 | 0.734 | 0.792 |