yaqiliu-cs / SelfDM-TIP

Released codes of Two-Stage Copy-Move Forgery Detection With Self Deep Matching and Proposal SuperGlue

SelfDM+PS

Citation

The code is for Research Use Only. If you use this code for your research, please consider citing:

@article{liuyaqi2022tip,
  title={Two-Stage Copy-Move Forgery Detection With Self Deep Matching and Proposal SuperGlue},
  author={Liu, Yaqi and Xia, Chao and Zhu, Xiaobin and Xu, Shengwei},
  journal={{IEEE} Transactions on Image Processing},
  volume={31},
  pages={541--555},
  year={2022},
  publisher={IEEE}
}

Acknowledgement

Our project relies on the codes from DeepMask, SuperGlue, and ConvCRF. You can download their pretrained models from their projects and kindly cite their papers.

Getting Started

The code was written and supported by Yaqi Liu.

Prerequisites

• Linux
• Python 3
• NVIDIA GPU + CUDA

Installation

conda create --name selfdm python=3.6 -y 
conda activate selfdm
conda install pytorch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 cudatoolkit=11.0 -c pytorch
pip install -r requirements.txt 

• Clone this repo:

git clone https://github.com/yaqiliu-cs/SelfDM-TIP
cd SelfDM-TIP

Generated BESTI dataset

• BESTI.

Pre-trained model

Download our pre-trained selfdm

• Test selfdm+PS model

python proposals_coco.py

Authorship

Yaqi Liu

Beijing Electronic Science and Technology Institute, Beijing 100070, China

E-mail: [e-mail]

Yifan Zhang

Beijing Electronic Science and Technology Institute, Beijing 100070, China

E-mail: [e-mail]