FORGERY SPOTTER SYSTEM USING HYBRID CNN–GAN ARCHITECTURE FOR DIGITAL IMAGE FORGERY DETECTION

Abstract

With the rapid growth of digital media and image editing tools, ensuring the authenticity of images has become a major challenge in journalism, digital forensics, cybersecurity, and social media. This research proposes a hybrid deep learning–based image forgery detection system that integrates Convolutional Neural Networks (CNN) and Generative Adversarial Networks (GAN) to detect copy-move and splicing forgeries. The model uses VGG16 for feature extraction and a GAN framework for generating realistic forged images to improve training diversity and detection robustness. The system is trained and evaluated using the CoMoFoD dataset, which contains both genuine and tampered images. Experimental results demonstrate improved accuracy, robustness, and adaptability compared with traditional and standalone deep learning approaches. The proposed system contributes to automated image authentication and helps combat misinformation in the digital era.