Analyzing Deep Learning Models’ Generalization Ability under Different Augmentations on Deepfake Datasets
Deepfakes allow users to manipulate identity of a person in a video or an image. Improvements on GAN-based
techniques also generate more realistic and harder to detect fake faces. This threatens individuals and decreases trust to
social media platforms. In this work our goal is to report four different models’ learning ability on, by far, largest fake face
dataset -DFDC and test the generalization ability of different models trained with this dataset and tested with Celeb-DF-v2.
Keywords - Deepfake, dfdc, Face Manipulation