/deep-splicing

Code related to my ICPR2016 paper about deep learning approaches to detect forged images

Primary LanguagePython

Deep Splicing

This code is used to apply Deep Learning framework for image foresnics (splicing detection)


Dependencies

Notes

Refer to config.ini for the configuration parameters and paths.

How To

CASIA_edge_exractor.m

It is a MATLAB script that exploits the P.Dollar edge detection toolbox to generate edges images. This procedure is needed in order to create the dataset.

The dataset has to be placed in several directories:

  • Authentics: containing the Authentics images of the dataset
  • Tampered: containing the Tampered images of the dataset
  • Au_Borders: containing the border extraction for Authentic images
  • Tp_Borders: containing the border extraction for Tampered images
  • Masks: containing the tampering masks generated automatically using the procedure in the paper

CASIA_training_test_creator.py

This is used to randomly generate the textual file containing the paths to the files for testing and training.

deep_tester.py

Runs the experiment.