Overview
FaceForensics++ is a comprehensive dataset designed for training and evaluating algorithms for detecting manipulated facial images and videos. It includes 1000 original video sequences sourced from YouTube, manipulated using four different automated face manipulation methods: Deepfakes, Face2Face, FaceSwap, and NeuralTextures. The dataset provides binary masks for segmentation tasks, making it suitable for both image and video classification. It also offers 1000 Deepfakes models for generating and augmenting new data. The Deep Fake Detection Dataset, provided by Google & JigSaw, is also included. The dataset is valuable for research in media forensics, security, and combating the spread of misinformation through manipulated media.