RAN: A randomness-anchored watermark attacking network with stealth and effectiveness
Abstract Research on watermark attacking is essential to reinforce robust watermarking methods by providing new attacking benchmarks. Recently, there is an emergence of attacking methods based on deep learning, in which perceptual loss and watermark loss are utilized to train the neural networks for...
Salvato in:
| Autori principali: | , , , , , , |
|---|---|
| Natura: | Artigo |
| Lingua: | Inglês |
| Pubblicazione: |
Nature Portfolio
2026-05-01
|
| Serie: | Scientific Reports |
| Soggetti: | |
| Accesso online: | https://doi.org/10.1038/s41598-026-52298-w |
| Tags: |
Nessun Tag, puoi essere il primo ad aggiungerne!!
|