Facial Recognition Software Trying to Catch Up With Face Masks
Facial recognition technologies seem to follow us anywhere we go and it is likely that won't be changing any time soon. Amid the COVID-19 pandemic, however, it has become evident that the face masks that we wear to protect each other provide a certain level of protection against facial recognition too, and a preliminary study from the National Institute of Standards and Technology (NIST) has taken a look at how they fare up most recently.
Not surprisingly, the facial recognition algorithms are having a hard time recognizing masked faces. However, that is not to say that this will be the case for a long time.
89 facial recognition algorithms including those by Panasonic and Samsung were analyzed
The study examined 89 facial recognition algorithms by looking at their performance on images of 1 million people.
When the photos were "digitally masked" by covering the subject's nose, mouth, and cheeks, the algorithms struggled. The results indicated that the presence of a mask reduced accuracy.
The press release read, "The best of the 89 commercial facial recognition algorithms tested had error rates between 5% and 50% in matching digitally applied face masks with photos of the same person without a mask."
The shape and color of the mask was an important factor too --the algorithms worked better with people wearing round masks. Also, black masks caused less accurate results than light-colored masks.
More developments will be made
While these might sound like uplifting news for those against such technologies, it seems like such errors might be temporary. Technology fails only mean that there are a lot of people who are trying to fix that issue.
This might mean that the pandemic might make facial recognition systems even powerful than before.
Albert Fox Cahn, the executive director of the Surveillance Technology Oversight Project, told Recode, "The good news here is very short-lived. This just highlights that there’s a global arms race right now to develop facial recognition software that can track people, even when we are wearing masks.”
The NIST study also touched on the issue by stating the struggles might be temporary. Computer scientist Mei Ngan who is one of the authors of the report said that the researchers are expecting the technology to improve and that they plan to use updated algorithms in other research.
It should also be noted that independent researchers are building databases by using photos of mask-wearing people. Such developments will only further improve the performance of facial recognition systems.