Source and Target Region Discrimination on Copy-Move Image Forgery Using SegFormer Model

Copy-move image forgery is a type of image manipulation where a part of an image is copied and pasted onto another part of the same image. Detecting copy-move image forgery involves identifying duplicated region and discriminating between the source and target regions, which can provide investigators with insights into the purpose of the forgery. This research aims to discriminate between the source and target regions in copy-move image forgery using SegFormer. SegFormer is a transformer-based semantic segmentation model. This research train SegFormer architecture from scratch to differentiate source and target regions. Experiments on the CoMoFoD dataset show that the SegFormer model achieved a mIoU score over 60%, demonstrating its effectiveness in discriminating source and target regions on a publicly available dataset.