CIDS

The official code for 'On Cover Independent Deep Neural Network Steganography'

• Python 3.8.13, PyTorch = 1.11.0

• Run the following commands in your terminal:

  conda env create -f env.yml

  conda activate NIPS

• Generating cover image set

  Run python stable_diffusion.py

• training dataset:

  DIV2K training dataset

• testing datasets:

1. DIV2K testing dataset
2. 1000 images randomly selected from the COCO dataset
3. 1000 images randomly selected from the CelebA dataset

1. Change the code in config.py

   line 4: mode = 'train'

2. Run python CIDS.py

1. Change the code in config.py

   line4: mode = 'test'

   line 41: test_cids_path = ''

2. Run python CIDS.py

• Here we provide trained models.
• The processed images, such as stego image and recovered secret image, will be saved at 'results/images'
• The training or testing log will be saved at 'results/cids_trained_on_div2k.log'

If you find our paper or code useful for your research, please cite:

@ARTICLE{11314630,
  author={Li, Guobiao and Li, Sheng and Luo, Zicong and Qian, Zhenxing and Zhang, Xinpeng},
  journal={IEEE Transactions on Dependable and Secure Computing}, 
  title={On Cover Independent Deep Neural Network Steganography}, 
  year={2025},
  volume={},
  number={},
  pages={1-16}
}