The DeepFake Detection Challenge Dataset

Brian Dolhansky, Joanna Bitton, Ben Pflaum, Jikuo Lu, Russ Howes, Menglin Wang, Cristian Canton-­Ferrer

arXiv (Cornell University) · 2020 · 185 citations · 2 references

Full text

Open access

Concepts

TL;DR

Deepfakes and other GAN‑based face‑swapping methods enable identity swapping in videos, and code for these techniques is publicly available. The authors built the DFDC dataset and organized a Kaggle challenge to train and evaluate deep‑fake detection models, and they also analyze top contest submissions. The DFDC dataset contains over 100,000 face‑swap clips from 3,426 actors, produced with multiple Deepfake, GAN‑based, and non‑learned methods, with all subjects consenting, and the data is publicly downloadable for training, validation, and testing. Models trained solely on DFDC generalize to real in‑the‑wild deepfakes, demonstrating that DFDC can support effective deep‑fake detection despite the problem’s difficulty.

Abstract

Deepfakes are a recent off-the-shelf manipulation technique that allows anyone to swap two identities in a single video. In addition to Deepfakes, a variety of GAN-based face swapping methods have also been published with accompanying code. To counter this emerging threat, we have constructed an extremely large face swap video dataset to enable the training of detection models, and organized the accompanying DeepFake Detection Challenge (DFDC) Kaggle competition. Importantly, all recorded subjects agreed to participate in and have their likenesses modified during the construction of the face-swapped dataset. The DFDC dataset is by far the largest currently and publicly available face swap video dataset, with over 100,000 total clips sourced from 3,426 paid actors, produced with several Deepfake, GAN-based, and non-learned methods. In addition to describing the methods used to construct the dataset, we provide a detailed analysis of the top submissions from the Kaggle contest. We show although Deepfake detection is extremely difficult and still an unsolved problem, a Deepfake detection model trained only on the DFDC can generalize to real in-the-wild Deepfake videos, and such a model can be a valuable analysis tool when analyzing potentially Deepfaked videos. Training, validation and testing corpuses can be downloaded from this https URL.

References

2