TensorFlow 2.5.0 Implementation of "Convolutional Neural Network-Based Image Watermarking using Discrete Wavelet Transform"
- Install Python packages
pip install -r requirements.txt
-
Download COCO dataset. Add train, validation, and test images to directories
./train_images
,./validation
, and./test_images
respectively. You can change the paths at theconfigs.py
file. -
Set the models output path at the
configs.py
file.
To run training:
python trainer.py
By default, it will save a model checkpoint every epoch to MODEL_OUTPUT_PATH
.
For more arguments and options, see configs.py
.
A notebook prepared for evaluation. To run the jupyter notebook use the script bellow:
jupyter notebook
After accessing notebook, open the evaluator.ipynb
and run the desire cells.
If you use this code as part of any published research, please refer the following paper.
@article{tavakoli2023convolutional,
title={Convolutional neural network-based image watermarking using discrete wavelet transform},
author={Tavakoli, Alireza and Honjani, Zahra and Sajedi, Hedieh},
journal={International Journal of Information Technology},
year={2023},
publisher={Springer}
}