Face Masks Disrupt Facial Recognition Systems

By The Biz Team (Nicholas)



The age of Covid-19 has made social interaction vastly more complicated. But it has also had some affects on the security scene, as the wearing of face masks has been normalized. Face masks are now a requirement in most public establishments, and have not only become a statement of safety, but fashion. The US Center for Disease Control and Prevention has repeatedly promoted the wearing of face coverings to prevent the spread of the virus, as the death toll approaches 170,000 in the United States alone. However safe, these masks have become a problem for agencies trying to identify security threats within the nation’s borders. 


Before the advent of the virus, providers of facial recognition were expected to install their technology in casinos, restaurants, airports and schools. Since the virus brought the necessity of face masks, it changed the industry's plan, though they are considering the situation to be more of a speed bump than a complete roadblock.


 The Department of Homeland Security is one of the prominent institutions which have expressed concerns on the use of face masks. A bulletin from the 22nd of May which was drafted in conjunction with other federal institutions such as Immigration and Customs Enforcement, discussed the potential effects that the use of masks might have on security activities that incorporate face recognition. Image processing hardware is an example where masks are an issue. Homeland security suggested in the memo that protesters could utilize the masks to avoid detection, though they did acknowledge that they did not have any specific information regarding violent extremists using protective face masks for attacks.


Since the CDC has encouraged the use of face coverings, cities across the country have begun reigning in the use of facial recognition technology by law enforcement and other entities. Boston, for example, banned the use of facial recognition during the last month, after areas of Oakland and San Francisco took similar action. The American Civil Liberties Union in Michigan filed a complaint against the law enforcement authority in Detroit, concerning a wrongful arrest of a man that was misidentified through the use of facial recognition.


According to Salon news, the bulletin admitted the agency does not have specific information that extremists or others in the country are using face coverings to execute attacks. However, there was speculation that such an incident could happen because the entities had expressed interest in circumventing face recognition, and promulgated instructions that allowed their cell members to hide their identity.


One example would be a member of a Middle Eastern extremist group that proposed the wearing of a breathing mask to hide a person’s identity, while also attacking the targeted infrastructure. In the summary statement, the agency showed a lot of confidence that extremists and other criminals would retain an interest in the avoidance of facial recognition systems, particularly in public areas. They only expressed medium confidence that the facial recognition processes would be effective during the public safety measures in the time of the Covid-19 emergency. They indicated that confidence would be increased based on reporting, which showed that extremists are particularly aware concerning the limitations of facial recognition. 


The problem is that law enforcement has become mired in controversy and concerns about civil liberties. To the public at this time, facial recognition is seen as something that infringes on individual privacy rights. Similarly, there are increased claims that algorithms used by the technology are racially biased, because they may mischaracterize people from certain ethnicities. However, they may show a higher accuracy when it comes to identifying those who are light-skinned. Due to external pressure, Amazon for example, announced that it would have a one year break on allowing the police department to use the facial recognition technology, depicted as Rekognition. From the public perspective, citizens need to wear masks during the pandemic, but the security agencies believe there is a safety cost incurred in the process. 


In response to the face mask phenomenon, several facial recognition companies have indicated that their technology can still identify people. Facewatch stated that they are releasing an algorithm that can handle detection and identification according to the eye and eyebrow area of a person. The company has proposed the technology to retail stores and stated that the development is going to extend beyond masks to other coverings, including hijabs which are worn by women of the Muslim faith. Facewatch had previously been working on identifying people that are wearing glasses and hats. The customers, who were mostly from retail stores looking to root out shoplifters, did not consider that mask detection would be a concern up until earlier this year. Now the technology is proving to be invaluable in identifying people, considering the need for face coverings is not going to deteriorate any time soon, and security agencies may quickly turn to companies such as Facewatch. They would design applications that identify people based on other attributes such as eyes, eyebrows, hairline and facial structure when the mask is considered.