Detection software is used to identify faces from video footage taken during the insurrection on January 6.
Photo Courtesy: Wired; Getty Images
Earlier this month, a website designed to help identify people at the Jan. 6 Capitol Hill riots went live, displaying thousands of facial images of people who allegedly participated in the fatal siege.
The site, titled Faces of the Riots features almost 6,000 images extracted from videos uploaded to Parler, a far-right social media platform which became popular for white supremacist content and users. Parler was recently shut down for condoning posts about the violent insurrection that resulted in the deaths of five people.
Not long before Parler was taken offline, hackers spent time archiving images and location data for images and videos that were uploaded to the fringe site. The creators of Faces of the Riots used open-source facial detection software to take images from over 800 videos that were posted on Parler from outside and within the US Capitol building.
The creator of the site, who wishes to remain anonymous, claims that his intent is to help the FBI identify who participated in the siege. Of course, this has garnered criticism from conservative groups and human rights leaders alike.
“Everybody who is participating in this violence, what really amounts to an insurrection, should be held accountable,” said the creator of the site.
Despite its original intent, Faces of the Riot has garnered quite a bit of criticism from conservatives and other groups. Many agree that Faces of the Riot infringes on the privacy of individuals and represents the pervasiveness of facial recognition technology.
Evan Greer, the campaign director for Fight for the Future, a digital civil liberties nonprofit believes that facial recognition technology has serious implications for “human rights and freedom of expression” whether it’s utilized by regular people or the government. In fact, Greer’s organization has lobbied for a legislative ban on facial recognition technologies.
The developer of the site disputes that Faces of the Riot simply relies on facial recognition but instead uses facial detection. He states that he used an open source machine learning tool called TensorFlow and recognition software called Dlib to analyze the videos from Parler. He says that he only used the software to detect faces from the videos and “deduplicated” nearly 200,000 facial images from the video frames to extract 6,000 specific faces. The site’s creator saw the data as an opportunity for a public project to aid law enforcement in finding those that stormed the Capitol building.
Despite the controversy behind the site’s methods, facial recognition and detection continues to be a prevalent way that law enforcement catches many criminals. Read more about the controversies behind facial recognition in the Spring 2021 issue of the Mace & Crown set to release in February 2021.