Department of Computer Engineering
Deepfake Detection Exploiting Subtle Cues
(Supervisor: Asst. Prof. Dr. Hamdi Dibeklioğlu)
Computer Engineering Department
Detection of manipulated videos and images has become a major field of research in computer vision and deep learning. With the recent improvement of Generative Adversarial Networks (GAN), there has been a surge of manipulated and synthesized videos online. Though mostly used for entertainment purposes, such content has proven to be potentially very damaging to individuals, organizations, and countries. Detection of manipulated content is increasingly important in a time of exponential information spread, which is why several major companies such as Google, Facebook and etc., have been heavily invested in similar research. We propose a learning-based Deepfake detection model with CNN+RNN hybrid architecture. We will be using The Deepfake Detection Challenge (DFDC), Celeb-DF, and FaceForensics++ for the training of the model in the initial phase of the research.
DATE: 18 November 2020, Wednesday @ 13:15