Abstract:
This paper proposes FakeBuster, a novel DeepFake detector for (a)
detecting impostors during video conferencing, and (b) manipulated
faces on social media. FakeBuster is a standalone deep learningbased solution, which enables a user to detect if another person’s
video is manipulated or spoofed during a video conference-based
meeting. This tool is independent of video conferencing solutions
and has been tested with Zoom and Skype applications. It employs
a 3D convolutional neural network for predicting video fakeness.
The network is trained on a combination of datasets such as Deeperforensics, DFDC, VoxCeleb, and deepfake videos created using
locally captured images (specific to video conferencing scenarios).
Diversity in the training data makes FakeBuster robust to multiple
environments and facial manipulations, thereby making it generalizable and ecologically valid
