Lightweight DL Model Based Detection of Image Forgery for Copyright Application

Abstract

In this digital era, images and videos are being used as influential sources of evidence in a variety of contexts like evidence during trials, insurance fraud, social networking, etc. The easy adaptability of editing tools for digital images, especially without any visual proof of manipulation, give rise to questions about their authenticity. It is the job of image forensics authorities to develop technological innovations that would detect the forgeries of images. There are three primary classes of manipulation or forgery detectors studies until now: those supported features descriptors, those supported inconsistent shadows and eventually those supported double JPEG compression. Image forgery detection is one of the key challenges in various real time applications, social media and online information platforms. The conventional methods of detection based on the traces of image manipulations are limited to the scope of predefined assumptions like hand-crafted features, size and contrast. In this paper, we propose a fusion based decision approach for image forgery detection. The fusion of decision is based on the lightweight deep learning models namely SqueezeNet, MobileNetV2 and ShuffleNet. The fusion decision system is implemented in two phases. First, the pretrained weights of the lightweight deep learning models are used to evaluate the forgery of the images. Secondly, the fine-tuned weights are used to compare the results of the forgery of the images with the pre-trained models. The experimental results suggest that the fusion based decision approach achieves better accuracy as compared to the state-of-the-art approaches.