Abstract
In this paper, we present our on-going effort of constructing a large-scalebenchmark, DeeperForensics-1.0, for face forgery detection. Our benchmarkrepresents the largest face forgery detection dataset by far, with 60, 000videos constituted by a total of 17.6 million frames, 10 times larger thanexisting datasets of the same kind. Extensive real-world perturbations areapplied to obtain a more challenging benchmark of larger scale and higherdiversity. All source videos in DeeperForensics-1.0 are carefully collected,and fake videos are generated by a newly proposed end-to-end face swappingframework. The quality of generated videos outperforms those in existingdatasets, validated by user studies. The benchmark features a hidden test set,which contains manipulated videos achieving high deceptive scores in humanevaluations. We further contribute a comprehensive study that evaluates fiverepresentative detection baselines and make a thorough analysis of differentsettings. We believe this dataset will contribute to real-world face forgerydetection research.