🌟RMSteg

Implementation for the paper Robust Message Embedding via Attention Flow-Based Steganography.

Our approach allows for message embedding into image with strong robustness against real-world image distortions (printing, JPEG compression, etc.).

📸 News

[Feburary 27, 2025] - Our paper is accepted at CVPR 2025!

[April 8, 2025] - Source code is released!

[April 16, 2025] - Pre-trained model is available!

🔧 Quick Start

From Pre-trained Model

You can download our pre-trained mode from here and place it in the src/pretrained/ folder. Then, you can run the following command for a quick start:

cd src
python single_test.py