Research Article
Image-to-Image Translation Generative Adversarial Networks for Video Source Camera Falsification
@INPROCEEDINGS{10.1007/978-3-031-36574-4_1, author={Maryna Veksler and Clara Caspard and Kemal Akkaya}, title={Image-to-Image Translation Generative Adversarial Networks for Video Source Camera Falsification}, proceedings={Digital Forensics and Cyber Crime. 13th EAI International Conference, ICDF2C 2022, Boston, MA, November 16-18, 2022, Proceedings}, proceedings_a={ICDF2C}, year={2023}, month={7}, keywords={Generative Adversarial Networks (GANs) Multimedia forensics Video Source Identification Machine learning}, doi={10.1007/978-3-031-36574-4_1} }- Maryna Veksler
Clara Caspard
Kemal Akkaya
Year: 2023
Image-to-Image Translation Generative Adversarial Networks for Video Source Camera Falsification
ICDF2C
Springer
DOI: 10.1007/978-3-031-36574-4_1
Abstract
The emerging usage of multimedia devices led to a burst in criminal cases where digital forensics investigations are needed. This necessitate development of accurate digital forensic techniques which require not only the confirmation of the data integrity but also the verification of its origin source. To this end, machine and/or deep learning techniques are widely being employed within forensics tools. Nevertheless, while these techniques became an efficient tool for the forensic investigators, they also provided the attackers with novel methods for the data and source falsification. In this paper, we propose a simple and effective anti-forensics attack that uses generative adversarial networks (GANs) to compromise the video’s camera source traces. In our approach, we adopt the popular image-to-image translation GANs to fool the existing algorithms for video source camera identification. Our experimental results demonstrate that the proposed attack can be implemented to successfully compromise the existing forensic methods with 100% probability for non-flat videos while producing the high quality content. The results indicate the need for attack-prone video source camera identification forensics approaches.
