Image and Video Forensics using Deep Learning

Abstract

Digital forensics is a critical research branch that focuses on identifying the original source and verifying the authenticity of digital data, particularly concerning visual information, which poses a major challenge for forensic experts in the present digital era. This field extensively employs image and video data to validate the accuracy of information during the picture gathering phase and to detect tampering throughout the digital image processing pipeline. However, image and video forensics stand as the most significant challenge in the current digital landscape due to the continuous modification of digital content using freely available software and editing tools. Ensuring the authenticity of images and videos involves employing various forensic methods from image acquisition to storage. Despite the complexity of this task, two major hurdles faced by researchers in digital forensics include identifying the source of acquired images and videos to establish their origins with specific devices, models, or brands, which holds particular importance when examining historical context. Additionally, detecting image forgery is essential to preserve the integrity of digital data for legal purposes, as forged images distort original content and disseminate false information. With the ready availability of software editing tools, the risk of modifying captured images with deceptive or defamatory content has increased, leading to the spread of fake information on social media platforms.