Embedding Guided End-to-End Framework for Robust Image Watermarking

Figure 7

Architecture of the discriminator. The numbers in the form m/n represent the kernel size (m) and the number of kernels (n) in each convolution layer.