Detection of Digital Image Forgery Using Deep-Learning

In contemporary times, digital images serve as a primary medium for disseminating information across social media platforms.However, malicious software can manipulate these images to spread false information.So, it's crucial to identify these forgeries.The literature addresses this issue through a range of digital image forgery detection techniques.However, most of these techniques are limited to detecting only one type of forgery, such as image splicing or copy-move, which is not practical for real-life applications.This paper presents an approach to improve digital image forgery detection by utilizing deep learning techniques through transfer learning, aiming to simultaneously uncover two types of image forgery.The proposed technique relies on discovering the compressed quality of the forged area, which normally differs from the compressed quality of the rest of the image.A deep learning-based model is proposed for detecting digital image forgery by calculating the difference between the original image and its compressed version.This process generates a featured image, which serves as input to the pre-trained model.The model is then trained by removing its original classifier and adding a new, fine-tuned classifier.A comparison of eight different pre-trained models adapted for binary classification has been conducted.The experimental results indicate that utilizing this technique with eight different adapted pre-trained models surpasses state-of-the-art methods.This conclusion is based on a comparison of the resulting evaluation metrics, charts, and graphs.Additionally, the results indicate that employing the technique with the pre-trained MobileNetV2 model achieves the highest detection accuracy rate (approximately 95%) with fewer training parameters, resulting in faster training times.