Selfies with a face mask are not only seen by your friends and family - they are also collected by researchers who want to use them to improve facial recognition algorithms. CNET found thousands of selfies with a mask for download on public databases, with photos that were "collected" directly from Instagram.
The COVID-19 pandemic has forced all people to wear face masks. Of course, this has created a problem for face recognition companies. Facial masks cover a significant part of what facial recognition needs to be identified and located people - essentially threatening the future of a multi-million dollar industry, unless technology can learn to recognize people without being influenced by the mask.
To do this, they need more covered photos to train their algorithms.
In April, researchers published the COVID19 mask data set on Github, using more than 1.200 images collected by Instagram. A month earlier, researchers from China compiled a database of more than 5.000 photos with a mask they collected online.
The creators behind the April database used Workaround AI startup to help identify the faces in the photos, with or without a mask, said Wafaa Arbash, the company's CEO.
“We were inspired by all the companies that were released for free tools and whatever they can do to help, "said Arbash. "It simply came to our notice then pictures from Instagram, so they are not private images. We just searched and found the right data. ”
Face recognition companies have long used images of people without their consent to train their algorithms. Proponents of civil liberties argue that face recognition technology threatens it private life and freedom of speech, also warning that there is almost no law that prevents the abuse of surveillance tools.
Clearview AI, a controversial face recognition company, has claimed the right to cut more than 3 billion images from social media for use on its database.
Governments in more than half of their states USA oblige citizens to wear masks in public because they help prevent the spread of COVID-19. The masks have also slowed the spread of facial recognition, as clothes block the main parts of your face that are usually analyzed by technology.
Some face recognition providers have turned to their own employees to send a selfie with a face mask to process the recognition process.
However, edited face mask photos may not be as effective as organic body painting images. algorithms. Face recognition companies also need a different set of images so that algorithms can better identify women, people of different ages and a variety of mask types.
For her company's public database, Arbash said the photos came from an Instagram search with hashtags associated with masks. They collected about 3.000 photos from the social media platform, but reduced them to a total of 1.200 photos. The published image samples included one photo of a child - Arbash said it was a possible mistake that this image ended up in the company's database.
Arbash said they did not ask people involved in the database for permission to use their selfies and that if they wanted to be excluded, they could make their pages private. The people involved are unaware that they are in this data set, he said.
"We don't make money from it, it's not commercial," Arbash said. "The goal and the intention was to help all the engineers in science data or the machine learning they are working on to fix this issue and help with public safety. ”