Facebook’s Artificial Intelligence researchers have developed an AI system that can “de-identify” a person in a recorded video and even on live videos in real-time. VentureBeat reports that this system uses machine learning to distort a person’s face in a video, making it difficult for the facial recognition system to identify.[wpinsertshortcodead id=”zxikm5d3e04a8f1451″]
A paper explaining this reads:
Face recognition can lead to loss of privacy, and face replacement technology may be misused to create misleading videos,” a paper explaining the approach reads. Recent world events concerning the advances in and abuse of face recognition technology invoke the need to understand methods that successfully deal with de-identification. Our contribution is the only one suitable for video, including live video, and presents quality that far surpasses the literature methods.
In a telephonic interview with VentureBeat, Facebook AI Research engineer and Tel Aviv University professor Lior Wolf said that Facebook’s approach pairs an adversarial autoencoder with a classifier network to distort a person’s face in such a way that it fools the facial recognition system. You can check the Jennifer Lawrence’s face and other celebrity faces generated with the AI system in the below video.
A Facebook spokesperson told VentureBeat that they have no plans to implement this in any of its products.
It is planned to present this work at ICCV (International Conference on Computer Vision) a few days later in Seoul, South Korea.