Skip to content

Official implementation of "Harmonizing Image Forgery Detection & Localization: Fusion of Complementary Approaches"

License

Notifications You must be signed in to change notification settings

IDLabMedia/fusion-idlab

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Fusion IDLab

Official implementation of "Harmonizing Image Forgery Detection & Localization: Fusion of Complementary Approaches"

Try-out without downloading or cloning

The fusion method is integrated in the COM-PRESS dashboard, where anyone can upload images for manipulation analysis. Or check out the interactive examples here.

Example of FusionIDLab on COM-PRESS dashboard

The COM-PRESS consortium consists of the research groups IDLab-MEDIA and mict (both from UGent-imec), and Apache (De Werktitel cv) as media partner. Moreover, the consortium is working with CheckFirst. The project received subsidies from the Flemish government’s Department of Culture, Youth & Media (Departement Cultuur Jeugd & Media).

COM-PRESS Consortium logos

Examples

Below, some visual examples of the heatmaps of the individual methods and the proposed fusion method are given. Examples of individual heatmaps and output of Fusion

Additionally, the examples_input and examples_output folders contain an example that is used in the Jupyter notebook run_forgery_detection_fusion.ipynb.

Dependencies

The code requires Python 3.X and was built with Tensorflow 2.15. Additionally, there are dependencies to two git submodules, comprint and CAT-Net, which require tensorflow and torch, respectively.

Install the requested libraries using:

pip install -r requirements_tf.txt
pip install -r requirements_torch_catnet.txt

Model weights

The fusion model weights can be downloaded by running the download_fusion_weights.sh script from the root fusion-idlab folder. The CAT-Net weights can be downloaded by running the download_weights.sh script from the CAT-Net folder. The comprint weights are included in the corresponding repository.

Usage (inference)

The Jupyter notebook run_forgery_detection_fusion.ipynb gives an example on how to run all individual methods, as well as the proposed fusion method. By adding new files in examples_input and changing the path in the notebook, you can extract the heatmaps from other images under investigation.

More information

More information can be found on our website.

The paper can be downloaded here.

Reference

This work was published in Journal of Imaging.

@Article{mareen2024fusion,
  AUTHOR = {Mareen, Hannes and De Neve, Louis and Lambert, Peter and Van Wallendael, Glenn},
  TITLE = {Harmonizing Image Forgery Detection & Localization: Fusion of Complementary Approaches},
  JOURNAL = {Journal of Imaging},
  VOLUME = {10},
  YEAR = {2024},
  NUMBER = {1},
  ARTICLE-NUMBER = {4},
  URL = {https://www.mdpi.com/2313-433X/10/1/4},
  PubMedID = {38248989},
  ISSN = {2313-433X},
  DOI = {10.3390/jimaging10010004}
}

About

Official implementation of "Harmonizing Image Forgery Detection & Localization: Fusion of Complementary Approaches"

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages