Passive Detection of Copy-Move Forgery Using Transfer Learning on Digital Image

Copy-move forgery is a common type of image tampering where a part of an image is copied and pasted elsewhere in the same image leaving a huge impact in evidence manipulation and spreading misinformation. Despite significant advancements, current research works are seen to take huge amount of time to train their model with large labeled dataset. In this research, a deep learning model is proposed for detecting copy-move forgeries. Transfer learning is used to train the model on two pre-trained CNN (Convolutional Neural Network) models, MobileNetV3 and EfficientNetV2. After training, the MobileNetV3 model achieved a test accuracy of 97.75%. This model can be used on low-storage devices or websites to detect copy-move forgeries and can be a help for forensic investigation, journalists, medical professionals & many others.