kylemin/WOUAF
WOUAF: Weight Modulation for User Attribution and Fingerprinting in Text-to-Image Diffusion Models (CVPR 2024)
This tool helps content creators and platforms embed a unique, hidden identifier into images generated by text-to-image AI models like Stable Diffusion. It takes your base text-to-image model and a specific user's fingerprint, then produces a modified model that generates images with that user's unique signature. Anyone who wants to trace the origin of AI-generated images back to a specific user will find this useful.
No commits in the last 6 months.
Use this if you need to hold users accountable for the content they generate using your text-to-image AI models by embedding a traceable signature.
Not ideal if your primary concern is generating images without any form of attribution or if you are not using diffusion-based text-to-image models.
Stars
26
Forks
7
Language
Jupyter Notebook
License
Apache-2.0
Category
Last pushed
Jun 14, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/kylemin/WOUAF"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
OPTML-Group/Unlearn-Saliency
[ICLR24 (Spotlight)] "SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in...
Shilin-LU/VINE
[ICLR 2025] "Robust Watermarking Using Generative Priors Against Image Editing: From...
WindVChen/DiffAttack
An unrestricted attack based on diffusion models that can achieve both good transferability and...
koninik/DiffusionPen
Official PyTorch Implementation of "DiffusionPen: Towards Controlling the Style of Handwritten...
Wuyxin/DISC
(ICML 2023) Discover and Cure: Concept-aware Mitigation of Spurious Correlation