Transferable Stealthy Adversarial Example Generation via Dual-latent Adaptive Diffusion for Facial Privacy Protection

The code of "Transferable Stealthy Adversarial Example Generation via Dual-latent Adaptive Diffusion for Facial Privacy Protection".

Setting up

Install dependency

pip install -r requirements.txt

Prepare model checkpoints and datasets

Download the model checkpoints and datasets here.

Run the code

python attack_DADP.py

Acknowledgments

This code builds heavily on DiffProtect and DiffAE. Thanks for open-sourcing!