Dataset for training and evaluating DiffVax: Optimization-Free Image Immunization Against Diffusion-Based Editing (ICLR 2026).
The DiffVax dataset is designed for training and evaluating image immunization models that protect images from diffusion-based editing attacks. The dataset contains high-quality images with corresponding masks and editing prompts, enabling research on optimization-free defense mechanisms against diffusion-based… See the full description on the dataset page:
https://huggingface.co/datasets/ozdentarikcan/DiffVaxDataset.
