NTUST Introduction to Information Security – Final Term Report (Spring 2025)
This repository contains the final course report for Introduction to Information Security at
National Taiwan University of Science and Technology (NTUST).
The report surveys recent techniques for securing diffusion-based generative models using watermarking and traceability strategies.
Course: NTUST Introduction to Information Security
File | Description |
---|---|
survey.pdf |
Final version of the term report, formatted with IEEE template |
- Diffusion models overview
- Latent-space watermark injection
- Pixel-space and steganographic methods
- Model fingerprinting and attribution techniques
- Legal & ethical challenges
- Benchmark criteria (robustness, fidelity, stealth)
Model-level Watermarking
├── Latent-space Watermarking
│ ├── CLUE-MARK (arXiv 2024)
│ ├── Robin (NeurIPS 2024)
│ └── Gauge Flow Matching, PT-Mark, etc.
└── Semantic-aware / Adversarial
└── Flexible Semantic WM (ACM 2024), PT-Mark
Output-level Traceability
├── Image/Video Watermarking
│ ├── CoSDA (AAAI 2025)
│ ├── DTR: Tree-Ring Video (ICASSP 2025)
│ └── NTU Editing WM, Stable-Diffusion
└── Decoder Inversion & Perturbation Defenses
├── Gradient-Free Decoder Inversion (NeurIPS 2024)
└── WM vs. Perturbation (OpenReview 2024)
Provenance Verification
├── Fingerprinting & Style Detection
│ ├── ICCV 2023 (Diffusion Signature)
│ ├── CVPR 2020, TPAMI 2023 (GAN Fingerprints)
│ └── Self-supervised Face Detection (OpenReview 2024)
├── Identity Embedding
│ └── Stable Signature (arXiv 2023)
└── Diffusion-aware Survey & Tabular Cases
├── Entropy 2025 (Spread-Spectrum WM)
└── TU Delft 2024 (Tabular Diffusion)
Ownership & Identity Protection
├── PCDiff: Proactive Control for Ownership (arXiv 2025)
└── WaDiff: Watermark-Conditioned Diffusion Model (arXiv 2024)
The following implementations cover key watermarking techniques across different architectures.
Category | Folder | Description |
---|---|---|
Latent-Space Watermarking | latent_space_watermarking (Robin code — unofficial) |
CLUE-MARK, Robin (Robin is for reference use only; not benchmarked) |
Pixel-Domain Watermarking | pixel_watermarking |
Direct image-space embedding (e.g., RivaGAN, DwtDct) |
Prompt-Guided Watermarking | prompt_guided_watermarking |
Custom zero-shot watermarking via prompt conditioning with CLIP-based verification |
Ownership Protection | None | Referenced method only (e.g., WaDiff (official repo) — not implemented here |
Provenance Verification | None | Referenced methods only (e.g., Stable Signature) — not implemented |
Each folder includes code and minimal examples.
For usage and execution instructions, please refer to the corresponding README.md
file inside each folder.
Title | Venue | Link | Technique | Code |
---|---|---|---|---|
CLUE-MARK: Watermarking Diffusion Models using CLWE | arXiv 2024 | arXiv:2411.11434 | Latent Watermarking | None |
PCDiff: Proactive Control for Ownership Protection | arXiv 2025 | arXiv:2504.11774 | Ownership + Traceability | None |
PT-Mark: Invisible Watermarking via Semantic Tuning | arXiv 2025 | arXiv:2504.10853 | Semantic-aware Tuning | None |
Towards a Correct Usage of Cryptography in Semantic Watermarks | arXiv 2025 | arXiv:2503.11404 | Cryptography-based Semantic Watermarking | None |
Robin: Robust and Invisible Watermarks for Diffusion Models | NeurIPS 2024 | Adversarial Watermarking | None | |
Invisible Yet Robust: Watermarking Diffusion Models with Adversarial Latents | arXiv 2024 | arXiv:2406.08337 | Latent Adversarial Watermarking | None |
Are Watermarks for Diffusion Models Radioactive? | OpenReview 2024 | Link | Watermark Robustness Analysis | None |
Gauge Flow Matching for Constrained Generative Modeling | ICLR Workshop 2025 | Constrained Diffusion | None | |
DiffusionGuard: Protecting Diffusion Models Against Visual Prompt Injection via Robust Watermarking | arXiv 2024 | arXiv:2410.05694 | Robust Output Watermarking | GitHub |
Title | Venue | Link | Technique | Code |
---|---|---|---|---|
CoSDA: Inversion-based Robust Watermarking | AAAI 2025 | Output Watermarking | None | |
DTR: Tree-Ring Watermarking for Videos | IEEE ICASSP 2025 | IEEE Abstract | Video Hierarchical WM | None |
Gradient-free Decoder Inversion in Latent Diffusion | NeurIPS 2024 | Link | Decoder Inversion | None |
Watermarks vs. Perturbations for Preventing Style Editing | OpenReview Workshop 2024 | Link | Defense Strategies | None |
Provenance Signature for Diffusion Model Outputs | CVPR 2024 | arXiv:2312.06688 | Output Provenance Tracking | None |
VIDiff: Video Diffusion Model Fingerprinting | CVPR 2024 | arXiv:2312.00286 | Video Output Fingerprinting | GitHub |
GenTrace: Provenance Tracing for Diffusion Models | ICLR 2024 | OpenReview | Latent-space Tracing | None |
DiffuseTrace: A Transparent and Flexible Watermarking Scheme for Latent Diffusion Model | arXiv 2024 | arXiv:2405.02696 | Semantic Latent Watermarking | None |
ROBIN: Robust and Invisible Watermarks for Diffusion Models with Adversarial Optimization | NeurIPS 2024 | arXiv:2411.03862 | Adversarial Watermarking | None |
ZoDiAc: Attack-Resilient Image Watermarking Using Stable Diffusion | NeurIPS 2024 | arXiv:2401.04247 | Latent Space Watermarking | GitHub |
InvisMark: Invisible and Robust Watermarking for AI-Generated Image Provenance | WACV 2025 | arXiv:2411.07795 | Neural Network-Based Watermarking | GitHub |
Shallow Diffuse: Robust and Invisible Watermarking through Low-Dimensional Subspaces in Diffusion Models | arXiv 2024 | arXiv:2410.21088 | Low-Dimensional Subspace Watermarking | None |
ProMark: Proactive Diffusion Watermarking for Causal Attribution | CVPR 2024 | Poster | Causal Attribution Watermarking | None |
Title | Venue | Link | Technique | Code |
---|---|---|---|---|
Self-supervised Representations for Detecting AI-Generated Faces | OpenReview 2024 | Link | Style-based Fingerprinting | None |
Who Made This Image? Detecting Generative Models via Traces in Diffusion | ICCV 2023 | arXiv | Diffusion Signature | None |
Stable Signature: Identity Watermarking for Stable Diffusion | arXiv 2023 | arXiv:2310.01856 | Watermark Embedding | None |
A Survey on GAN Fingerprints | IEEE TPAMI 2023 | DOI | Survey | None |
Attributing Fake Images to GANs: Learning and Analyzing Fingerprints | CVPR 2020 | CVF Link | Fingerprinting | None |
Watermarking Diffusion Models via Latent Space | NeurIPS 2024 | NeurIPS 2024 Paper Link | Latent-space Watermarking | None |
Provenance Tracing for Diffusion Models | CVPR 2024 | CVPR 2024 Paper Link | Provenance Tracing | None |
Secure Watermarking for Diffusion-Based Generative Models | ICLR 2024 | ICLR 2024 Paper Link | Secure Watermarking | None |
Title | Venue | Link | Technique | Code |
---|---|---|---|---|
PCDiff: Proactive Control for Ownership Protection | arXiv 2025 | arXiv:2504.11774 | Ownership Traceability | None |
A Watermark-Conditioned Diffusion Model for IP Protection | arXiv 2024 | arXiv:2403.10893 | Ownership Fingerprinting | GitHub |
TraceMark-LDM: Authenticatable Watermarking for Latent Diffusion Models | arXiv 2025 | arXiv:2503.23332 | Owner & User Identity Watermark | None |
Dynamic Watermarks in Images Generated by Diffusion Models | arXiv 2025 | arXiv:2502.08927 | QR-code Embedded Watermark | None |
GROOT: Generating Robust Watermarks for Diffusion-Model-Based Audio Synthesis | ACM MM 2024 | ACM Link | Ownership Traceability for Audio | None |
Watermarking for Stable Diffusion Models | IEEE IoT Journal 2024 | IEEE Link | Invisible Watermarking | None |
Diffusetrace: Transparent and Flexible Watermarking for Latent Diffusion Models | arXiv 2024 | arXiv:2405.02696 | Flexible Ownership Watermark | None |
Protect-your-IP: Scalable Source-Tracing Against Personalized Generation | arXiv 2024 | arXiv:2405.16596 | Source-Tracing Attribution | None |
Aqualora: Toward White-Box Protection via Watermark LoRA | arXiv 2024 | arXiv:2405.11135 | LoRA Watermarking | None |
Watermark-Embedded Adversarial Examples Against Diffusion Models | CVPR 2024 | CVPR Link | Visible Ownership Watermark | None |