Facebook De-Identification Technology
(Photo : Photo by Glen Carrie on Unsplash)

Facial recognition systems are currently all the frenzy among different government agencies worldwide as they seek out to automate services and keep accounts on their residents. Say, there's a photo of you somewhere. You can be spotted in videos and photos from the feeds of public cameras. At present, Facebook has developed a way to prevent this technology. 

The social media company's face de-identification, which three of its Artificial Intelligent (AI) researchers developed, can modify a person's face slightly in video content. The new technology is designed so that the facial recognition systems won't be able to match what comes out in the footage with your picture in their database. You can see this innovation in which some details are tweaked or modified like the shape of the mouth of a person, or the size of his eyes. According to Facebook, this can be employed with either a pre-recorded content or a live video.

De-Identification for Protection of Privacy

Face recognition can also result in loss of privacy, and the technology for face replacement may be used for the wrong purpose to create misleading videos. Apparently, it is not Facebook's intention to use the De-Identification technology "in any of its commercial products." However, the research may impact future tools developed to protect the privacy of individuals, as the study focuses on 'misleading videos,' stop someone's likeness from being used in the so-called 'video deepfakes.'

In addition, this latest technological development from Facebook can allow for more ethical employment of the video footage of individuals for training AI systems that typically need several examples to learn how to copy the content they are fed. By making it impossible to recognize people's faces, the new AI systems can be trained minus violating the privacy of the subject being tested.

How the De-Identification Technology Works

Online News explained how this face-modifying system works. It says the AI utilizes the encoder–decoder architecture generating both an image and a mask. During the training, an individual's face gets distorted and fed into the network. Following this, the system then generates the distorted and undistorted image of the person's face for a result that can be inserted into the video.

As earlier mentioned, online news site also reports, Facebook doesn't have any plans of using de-identification tech for promotion of its products. The social media network has a face recognition app to identify people for easy-tagging and alert when one is in someone else's photos. However, it recently turned off its facial recognition by default, indicating its strong support for the users' privacy.

Facebook has always been in the news lately as it gets involved in various issues, both good and bad. Recently, following many years of reproach from the media industry because of its approach to the sector, which caused the shutdown of many publications, Facebook launched News Tab. This relatively new technology would enable the social media platform to pay media firms for the news content it posts. 

ⓒ 2024 TECHTIMES.com All rights reserved. Do not reproduce without permission.
Join the Discussion