Code for "A Full-Image Full-Resolution End-to-End-Trainable CNN Framework for Image Forgery Detection" The paper can be found on arxiv.
Download the model weights in ./model/E2E folder from here
To run the code on given test dataset, type:
python test.py -m=FUSION
This will run the fusion of the three models proposed in the paper (E2E-FUSION) on images in ./test/ directory.
For running on a single modality, change 'mode' parameter: 'RGB'-'N'-'RGN'
To run the code on your own dataset, type:
python test.py --te=PATH_TO_YOUR_DATASET
Where PATH_TO_YOUR_DATASET is organized with two subfolder: '0' contains pristine images and '1' contains forged images.
Modify the test.py file according to your needs.
For any problem or comment, do not hesitate to contact me.
@misc{marra2019e2e,
title={A Full-Image Full-Resolution End-to-End-Trainable CNN Framework for Image Forgery Detection},
author={Francesco Marra, Diego Gragnaniello, Luisa Verdoliva, Giovanni Poggi},
year={2019},
eprint={1909.06751},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
Copyright (c) 2019 Image Processing Research Group of University Federico II of Naples ('GRIP-UNINA').
All rights reserved.
This software should be used, reproduced and modified only for informational and nonprofit purposes.
By downloading and/or using any of these files, you implicitly agree to all the terms of the license, as specified in the document LICENSE.txt (included in this package)