SDFVD is a video dataset comprised of 106 clips in total: 53 real (original) videos and 53 deepfake videos.
The original videos were sourced from Pexels (stock footage provider).
The videos cover a variety of backgrounds and subjects of different genders and ages (hence diverse contexts).
Each clip has been pre-processed to:
Duration: ~4 to 5 seconds
Resolution: 720p
Uniform format for consistency.
The deepfake videos were generated using Remaker AI (face-swap technique). Source face photos came from Freepik