The Deepfake Detection Challenge (DFDC) Preview Dataset

By Brian Dolhansky, Russ Howes, Ben Pflaum, Nicole Baram, Cristian Canton-Ferrer. 2019

663 citations recorded by the source corpus.

In this paper, we introduce a preview of the Deepfakes Detection Challenge (DFDC) dataset consisting of 5K videos featuring two facial modification algorithms. A data collection campaign has been carried out where participating actors have entered into an agreement to the use and manipulation of their likenesses in our creation of the dataset. Diversity in several axes (gender, skin-tone, age, etc.) has been considered and actors recorded videos with arbitrary backgrounds thus bringing visual variability. Finally, a set of specific metrics to evaluate the performance have been defined and two existing models for detecting deepfakes have been tested to provide a reference performance baseline. The DFDC dataset preview can be downloaded at: deepfakedetectionchallenge.ai

Subfields: Computer Vision and Pattern Recognition, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Safety, Robustness & Interpretability

Citations in the corpus

Imperfect ImaGANation: Implications of GANs Exacerbating Biases on Facial Data Augmentation and Snapchat Selfie Lenses (2020, background): “…models on facial data is already prevalent in critical decision-making scenarios such as employment (Hymas, 2019), healthcare (Bahrampour, 2014), education (Kaur & Marco, 2019), criminal justice (Harwell, 2019), as well as security innovations like deep-fake detection (Dolhansky et al., 2019).”
The DeepFake Detection Challenge Dataset (2020, background): “…Dataset is both the largest currently-available Deepfake dataset, and one of only a handful of datasets containing footage recorded speciﬁcally for use in machine learning tasks (the others being the much smaller Google Deepfake Detection Dataset [6] and the preview version of this dataset [5]).”
DeeperForensics-1.0: A Large-Scale Dataset for Real-World Face Forgery Detection (2020, methodology): “, FaceForensics++ [33], Deep Fake Detection [9] and DFDC [14]) comprising manipulated video footages.”
One Shot Face Swapping on Megapixels (2021, background): “Besides, it is time consuming and difficult for subject-specific methods to train specific models for distinct pairs of faces [14, 46, 45, 29, 20, 55, 26].”
Generalized Out-of-Distribution Detection: A Survey (2021, neutral): “image forensics [42], [43], [44], industrial inspection [45], [46], [47], etc.”
GANprintR: Improved Fakes and Evaluation of the State of the Art in Face Manipulation Detection (2019, result): “1) XceptionNet [44]: this network was selected, essentially because it provides the best detection results in the most recently published studies [9], [27], [45].”
Creating, Using, Misusing, and Detecting Deep Fakes (2022, background): “By comparison, the leading DFDC predictor correctly identified 80% of the deep-fake videos, although this accuracy is significantly higher than the some 65% accuracy achieved by the topperforming model on the entire DFDC holdout dataset of 4,000 videos.”
Celeb-DF: A Large-Scale Challenging Dataset for DeepFake Forensics (2019, methodology): “To date, we have the UADFV dataset [53], the DeepFake-TIMIT dataset (DF-TIMIT) [25], the FaceForenscics++ dataset (FF-DF) [40] 2 , the Google Deep-Fake detection dataset (DFD) [15], and the FaceBook Deep-Fake detection challenge (DFDC) dataset [14].”
DeepFakes and Beyond: A Survey of Face Manipulation and Fake Detection (2020, methodology): “In [51], in addition to the description of the database, the authors have provided base-line results using three simple detection systems: i) a small CNN model composed of 6 convolution layers and 1 fully-connected layer to detect low-level image manipulations, ii) a XceptionNet model trained using…”
CoReD: Generalizing Fake Media Detection with Continual Representation using Distillation (2021, methodology): “Recently, several more advanced deepfake datasets (DFDC [17], CelebDF [45], CelebFOM [64], and WildDeepfake [80]) as well as GAN image datasets (CelebBlend [64], GANsformer [25], and StytleGAN2 [34]) are released.”
Two-branch Recurrent Network for Isolating Deepfakes in Videos (2020, methodology): “Ablation studies and comparisons are conducted on (1) FaceForensics++ [50], (2) Celeb-DF [38], (3) and the Deepfake Detection Challenge (DFDC) Preview Dataset [14].”
WildDeepfake: A Challenging Real-World Dataset for Deepfake Detection (2020, methodology): “These fake videos have raised enormous public concerns for their huge risks to create political distress, blackmail someone or even fake terrorism events [9].”
Deepfakes generation and detection: state-of-the-art, open challenges, countermeasures, and way forward (2021, background): “Deepfake Detection Challenge (DFDC): Recently, the Facebook community launched a challenge, aptly named the Deepfake Detection Challenge (DFDC)-preview [295], and released a new dataset that contains 1131 original videos and 4119 manipulated videos.”
Deep learning for deepfakes creation and detection: A survey (2019, methodology): “Videos DeepfakeTIMIT [147] and DFDC [109].”
Detecting Deepfakes with Self-Blended Images (2022, methodology): “DeepFake Detection Challenge Preview [20] (DFDCP) and DeepFake Detection Challenge public test set [19] (DFDC), that are released along with the competition [3], contain a lot of disturbed videos, e.”
Local Relation Learning for Face Forgery Detection (2021, neutral): “, Deepfake Detection Challenge (DFDC) (Dolhansky et al. 2019), Celeb-DF (Li et al.”