Tree-Rings Watermarks: Invisible Fingerprints for Diffusion Images

Authors: Yuxin Wen, John Kirchenbauer, Jonas Geiping, Tom Goldstein

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We perform experiments on two common diffusion models to measure the efficacy and reliability of the Tree-Ring Watermarking technique across diverse attack scenarios. Furthermore, we carry out ablation studies to provide an in-depth exploration of this technique.
Researcher Affiliation Academia Yuxin Wen, John Kirchenbauer, Jonas Geiping, Tom Goldstein University of Maryland {ywen, jkirchen, jgeiping, tomg}@umd.edu
Pseudocode No The paper describes methods using prose and mathematical equations, but does not include any explicitly labeled pseudocode or algorithm blocks.
Open Source Code Yes Code is available at https://github.com/Yuxin Wen Rick/tree-ring-watermark.
Open Datasets Yes The FID of Stable Diffusion is evaluated on the MS-COCO-2017 training dataset [Lin et al., 2014], and the FID of the Image Net Model is gauged on the Image Net-1k training dataset [Deng et al., 2009].
Dataset Splits No The paper specifies the training datasets used for FID evaluation (MS-COCO-2017 and Image Net-1k) and the number of generated images, but it does not provide specific train/validation/test dataset splits or explicitly mention a validation set.
Hardware Specification Yes All experiments are conducted on a single NVIDIA RTX A4000.
Software Dependencies No The paper mentions using Stable Diffusion-v2 and scipy, but it does not specify version numbers for any software dependencies like Python, PyTorch, or specific libraries.
Experiment Setup Yes In the main experiment, we use 50 inference steps for generation and detection for both models. For Stable Diffusion, we use the default guidance scale of 7.5, and we use an empty prompt for DDIM inversion... The watermark radius r we use is 10.