1 paper
Fei Ge, Ying Huang, Jie Liu +4
Existing deep image watermarking methods follow a fixed embedding-distortion-extraction pipeline, where the embedder and extractor are weakly coupled through a final loss and optim…