Forgery Image Dataset

This is a dataset regarding the creation of forgery images. The dataset consists of 1000 original and 3000 forgery images generated from the original images. The original images have been retrieved from publicly available repositories. Three different models have been used for creating the forgery images: cut-paste, copy-move, and erase-filling. Both pre-processing (sharpening, color enhancement, resizing, blurring, regulating exposure) and post-processing (sampling, rotation, masking) techniques have been considered for the generation of the forgery images. The dataset can be used for research in the area of fake image identification. This dataset has been created in the context of the H2020 project SocialTruth. This dataset is not available from this repository, see DOI for full access