DeepFakes: a New Threat to Face Recognition? Assessment and Detection

By Pavel Korshunov, Sébastien Marcel. 2018

801 citations recorded by the source corpus.

It is becoming increasingly easy to automatically replace a face of one person in a video with the face of another person by using a pre-trained generative adversarial network (GAN). Recent public scandals, e.g., the faces of celebrities being swapped onto pornographic videos, call for automated ways to detect these Deepfake videos. To help developing such methods, in this paper, we present the first publicly available set of Deepfake videos generated from videos of VidTIMIT database. We used open source software based on GANs to create the Deepfakes, and we emphasize that training and blending parameters can significantly impact the quality of the resulted videos. To demonstrate this impact, we generated videos with low and high visual quality (320 videos each) using differently tuned parameter sets. We showed that the state of the art face recognition systems based on VGG and Facenet neural networks are vulnerable to Deepfake videos, with 85.62% and 95.00% false acceptance rates respectively, which means methods for detecting Deepfake videos are necessary. By considering several baseline approaches, we found that audio-visual approach based on lip-sync inconsistency detection was not able to distinguish Deepfake videos. The best performing method, which is based on visual quality metrics and is often used in presentation attack detection domain, resulted in 8.97% equal error rate on high quality Deepfakes. Our experiments demonstrate that GAN-generated Deepfake videos are challenging for both face recognition systems and existing detection methods, and the further development of face swapping technology will make it even more so.

Subfields: Computer Vision and Pattern Recognition, Generative Models, 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 Generative Models

In Safety, Robustness & Interpretability

Citations in the corpus

WoodScape: A Multi-Task, Multi-Camera Fisheye Dataset for Autonomous Driving (2019, background): “is also the possibility of using GAN based re-targeting wherein faces are exchanged by automatically generated ones [31].”
FakeAVCeleb: A Novel Audio-Video Multimodal Deepfake Dataset (2021, background): “There are several deepfake detection datasets proposed by different researchers [10, 15, 17, 19, 22, 44] that are available publicly.”
Multi-task Learning for Detecting and Segmenting Manipulated Facial Images and Videos (2019, methodology): “Most of them were evaluated using only one or a few databases, including the CGvsPhoto database [25], the Deepfakes databases [2, 16, 17], and the FaceForensics/FaceForensics++ databases [26, 27].”
Exposing Deep Fakes Using Inconsistent Head Poses (2018, methodology): “Neural network based classification approach has also been used to directly discern real imagery from Deep Fakes [12, 13, 14].”
Emotions Don't Lie: An Audio-Visual Deepfake Detection Method using Affective Cues (2020, methodology): “We perform experiments on the DF-TIMIT [33] and DFDC [23] datasets, as only these datasets contain modalities for face and speech features (1).”
Spatial-Phase Shallow Learning: Rethinking Face Forgery Detection in Frequency Domain (2021, background): “Under this setting, these approaches achieve excellent performance on some public datasets [49, 21, 38, 27, 46].”
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].”
Video Transformer for Deepfake Detection with Incremental Learning (2021, methodology): “[31] employ CNNs to detect face warping artifacts in images from the deepfake datasets [29, 55].”
WildDeepfake: A Challenging Real-World Dataset for Deepfake Detection (2020, methodology): “Particularly, I3D achieves a high detection accuracy of > 89% on the four DF-TIMIT and FF++ datasets, while C3D has an accuracy of 73.18% on DFD.”
Deepfake Video Detection Using Convolutional Vision Transformer (2021, background): “Currently, it is possible to seamlessly generate hyper-realistic digital images [28] with a little resource and an easy how-to-do instructions available online [30, 9].”
Deepfakes generation and detection: state-of-the-art, open challenges, countermeasures, and way forward (2021, methodology): “2 DeepfakeTIMIT DeepfakeTIMIT [158] is another standard dataset for deepfake detection which was introduced in 2018.”
DiffusionDB: A Large-scale Prompt Gallery Dataset for Text-to-Image Generative Models (2022, background): “Deepfakes refer to fake imagery of real individuals for unethical and malicious applications (Korshunov and Marcel, 2018).”
Limits of Deepfake Detection: A Robust Estimation Viewpoint (2019, background): “While several studies use features extracted based on visual artifacts, image quality, lipsync, blinking, or warping for classiﬁcation (and might soon be obsolete) [10], [11], [12], [13], this work gives a generalizable statistical framework with guarantees on its reliability.”
The DeepFake Detection Challenge Dataset (2020, methodology): “Furthermore, as we observed that many publicly-released datasets [14, 17, 18, 23, 30] did not guarantee that their subjects were willing participants or agreed to have their faces modified, we solicited video data from 3,426 paid actors and actresses speaking in a variety of settings for roughly 15 ”