Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
ketejun 85e6f0f610 | 1 year ago | |
---|---|---|
blurpool | 1 year ago | |
models | 1 year ago | |
pytorch-colors @ 88624e57da | 1 year ago | |
utils | 1 year ago | |
LICENSE | 1 year ago | |
README.md | 1 year ago | |
best_model.pth | 1 year ago | |
dice_loss.py | 1 year ago | |
generate_edge.py | 1 year ago | |
inference.py | 1 year ago | |
losses.py | 1 year ago | |
options.py | 1 year ago | |
requirements.txt | 1 year ago | |
run_train.sh | 1 year ago | |
test.jpg | 1 year ago | |
train.py | 1 year ago |
Official repository for "Image splicing forgery detection by combining synthetic adversarial networks and hybrid dense U‐net based on multiple spaces"
For SAN:
Merge CASIA v2.0 and Forensics, and put the combination dataset (1891 images) into the subdirs /img/
and /mask/
.
HDU-Net/
├── ...
└── SAN/
├── img/
└── mask/
Generate a dataset that I call $SF-Data$ (82608 images). You can download $SF-Data$ via the link https://drive.google.com/file/d/1IoG78dAcxyw5fRPo1DjisKUoykJQTs2_/view?usp=sharing.
For HDU-Net:
To generate edge information according to the subdir /mask/
, run
python SAN/generate_edge.py
For SAN:
If you want to retrain SAN, run
python SAN/train.py
For HDU-Net:
To train HDU-Net, run
python train.py
You should change diverse parameters in options.py
For SAN:
I provide a well-trained model weight best_model_for_SAN.pth
.
You can download the weights via https://drive.google.com/file/d/1Qbn3kCxwMA7r-VQ0mpnXaI1tKKetPn7n/view?usp=sharing, and put it into the subdir /SAN/
.
You can use it to generate dataset based on other datasets like COCO, etc. I have converted the multi-label annotations "train2017" in COCO to binary mask. I are hesitating to upload the dataset since it is too large.
After running the following codes, you should change the path of dataset in options_GAN.py
.
Note that the well-trained weight only accept binary mask.
python SAN/generate_data.py
For HDU-Net:
You can download the weights of HDU-Net via https://drive.google.com/file/d/1XDMZdGzxSvs22j5uKx6Ywm92JlgFv_iw/view?usp=sharing.
Run
python inference.py
If you find this project useful for your research, please use the following BibTeX entry.
@article{Wei2022ImageSF,
title={Image splicing forgery detection by combining synthetic adversarial networks and hybrid dense U‐net based on multiple spaces},
author={Yang Wei and Jianfeng Ma and Zhuzhu Wang and Bin Xiao and Wenying Zheng},
journal={International Journal of Intelligent Systems},
year={2022}
}
This project is released under the MIT license.
Contact yale ywei9395@gmail.com for any further information.
No Description
Python other
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》