Multimodal Systems for Deepfake Video Detection
The project aims on the development of an audio-visual deepfake detection system.
Project GitHub: Not Ready.
The project is conducted by my 2nd year student, Larin Ilya, at HSE (Coursework) in 24/25 academic year.
To ensure trust in the fake data, it might not be enough to falsify only one modality (audio, face image or video). Therefore, it is also important to develop a countermeasure system that can handle multimodal deepfakes. Ideally, it should still work well in cases if only audio or video part of the data is spoofed. In this project, the student’s goal is to develop such a system.