Face biometric systems promise protection with new capabilities, but effectiveness not yet proven
Biometric facial recognition with thermal detection systems continue to be rolled out by the industry to meet the expected requirements for reopening businesses around the world. Mask detection is being added to face biometrics systems, and algorithms are being update to perform identification of faces occluded by masks. But how well do they work? The question remains open, even as business and governments scramble to put systems in place to reopen the global economy without inadvertently causing a second wave of death and lockdowns.
Japanese firms have developed new technologies for COVID-19 containment, including facial recognition that works with masks and new touch-free sensors, the South China Morning Post reports.
Glory Ltd., which makes money-changing machines, has developed facial recognition for masked faces and is planning to offer it as part of an access control product.
eConnect has launched a new thermal imaging device with facial recognition for self-service, contactless temperature checks at casinos and other businesses.
The new eClear can be integrated with employee databases to perform time and attendance tracking, and permanently mounted or deployed on a tripod to employee or guest entrances. The system also performs real-time aggregate counts of people entering and exiting, and provides alerts on recognizing VIP or banned customers, even if they are wearing a mask, according to the announcement.
The company says it intends to provide services and flexibility to help customers deal with specific circumstances and new requirements related to COVID-19 on a webpage dedicated to solutions for reopening.
Parsons Corporation has developed an integrated suite of solutions for touchless real-time health monitoring for public areas. In addition to biometrics and body temperature detection, the DetectWise kiosk includes heart and respiration rate measurement.
The suite also includes modular testing and decontamination facilities, and is intended for high-traffic areas such as airports, shopping malls, corporate buildings and stadiums. The contactless, mobile health screening system is expected to be deployed across the nation over the next few weeks, according to the company announcement.
The company says modular facilities will be invaluable as testing capabilities are required outside of traditional healthcare settings, and those offered as part of DetectWise will follow methods approved by the Centers for Disease Control and Prevention (CDC).
IN Depth has also launched a Fever Detection Camera system, which the company says utilizes advanced AI and scene analytics to perform detection and assessments of elevated body temperature in real-time, in addition to facial recognition.
The system uses thermal imaging and visible sensors, and can scan up to 30 individuals per second at a distance, with temperature accuracy within 0.3 degrees. The facial recognition database can hold up to 20,000 faces for real-time identification, and mask-wearing and removal can also be detected. The system’s AI algorithms also perform facial characteristic analysis for gender, age, and expression estimation.
The new capabilities of these technologies are as of yet untested by third parties, and even by customers, in most cases, leaving organizations with the unenviable task of sorting through myriad claims and offerings without the usual resources to judge them.
Patrick Grother, biometrics testing lead for the U.S. National Institute of Standards and Technology (NIST) recently told Biometric Update that the agency is planning to test facial recognition systems against masked faces, but the timeline for such testing is still unknown.
New systems have already been deployed by governments, hospitals, and enterprises, and developed to detect violations of physical distancing requirements. Trueface has used synthetic facial images in order to build a dataset to evaluate its newly-developed mask detection capability, but the company’s CEO Shaun Moore recently noted to Wired that it will really only know how well the technology works in the field to identify masked faces after the new capabilities has been deployed.
For one-to-one facial recognition on mobile devices, VinAI has developed an algorithm it says identifies people wearing surgical masks, and work-arounds for FaceID have been proposed, even as training with new datasets is carried out.
Many access control systems could be upgraded with the new offerings, but for higher security applications, however, any trade-off in security for the convenience of keeping a mask on may be misguided.
“When accessing a device, losing a bit of security – given the security level is already pretty low – is not an issue. But when accessing a bank account or asking for entrance into a restricted room, it is imperative the security level stays in place,” FaceTec VP of Communications John Wojewidka, who also edits Liveness.com, tells Biometric Update in an email. “Dropping a mask for about 3-5 seconds is well worth the effort.
Wojewidka also notes that the company’s technology is predominantly used in private places, rather than public ones.
For now, knowing whether to deploy proven biometric technology or new capabilities, and how to judge between systems yet to be tested in live operations is just another set of challenges for organizations looking forward to a return to business.