Facial mask recognition algorithms failing new study shows - Facial recognition has become more complicated with the widespread use of face masks. Although the algorithms have been specifically trained to work with the plug-in, the errors of this technology have multiplied with its use.

This is stated by a study by the U.S. National Institute of Standards and Technology (NIST), the competent authority responsible for reviewing the accuracy of this and other technology. For months now, NIST has been measuring the effectiveness of this biometrics in relation to the arrival of masks.

Initially, they found that the lack of accuracy ranged from 5 to 50%, when testing was done with existing algorithms. And now, after the latest results, including new specific algorithms arising from the pandemic, facial recognition errors have increased by 99%.

Read more: I tried a facial recognition tool to see which photos of mine are posted on the internet, and the results have made me worry even more about my privacy

In this variety of cases, the specialized Chinese firm Dahua has stood out for the effectiveness of its formula. While without face protection its error level was at 0.3% , with its use it has reached 6%.

Facial mask recognition algorithms failing new study shows

But at the opposite end there are also examples. Rank One, another facial recognition company, has gone from failing 0.9% to failing 34.8%. That is, its algorithm does not work more than 1 out of 3 times.

Read more: a Spanish company has just announced that its facial recognition technology is able to identify its users even if they wear masks

NIST has measured the effectiveness of more than a hundred companies in this latest study, whose data will be updated periodically.

"The results will be updated monthly, each time the algorithms and calculations are completed, as new data sets are added And when we do new analysis," the report says.

End of Facial mask recognition algorithms failing new study shows


More on:

The meter-and-a-half-distance rule is based on science from 80 years ago: scientists at MIT and Oxford have created this alternative system

At the end of the 19th century, the German scientist Carl Fl├╝gge had a hunch: keeping enough distance between people who are sick and others who are healthy might prevent the spread of pathogens.

At that time it was just a hypothesis, one more of those scientists were trying to test on lab plates.

However, it took 4 decades before the technology advanced enough to confirm it, thanks to the development of photography.

In the early 1940s, scientists were able to see people's sneezing for the first time in real time, with a capture speed of 30,000 frames per second. This confirmed that the particles that we expel when coughing and sneezing are deposited in the ground at a distance of one meter or one and a half meters.

At that time, scientists came to the conclusion that most of the infectious particles that people expelled (90% of the pathogens) traveled a distance of less than one and a half meters.

The measures of his study never intended to become strict and quick rules about how far we should be from another person during a pandemic.

However, this one-and-a-half rule has become an easy-to-follow protocol for keeping potentially sick people at bay during the coronavirus outbreak.

"This is how dogma was born," says Professor Lidia Morawska, a renowned scientist at the Australian aerosol study. "And like all dogma, it is very difficult to eliminate."

But as the pandemic gets longer, Morawska and other colleagues are beginning to question the safety distance and think of another more nuanced way to prevent the spread of the virus.

Instead of always being alert or assuming that distance Security keeps us safe all the time, these experts argue that we should learn to assess day-to-day situations, lowering our guard when it is relatively safe and returning to be alert if necessary.

This Tuesday, scientists from Oxford and MIT have released a system that works like a traffic light in the hope of helping people eliminate outdated dogmas and live life to the fullest but with sufficient care during the pandemic.

"With easy-to-use knowledge and tools, our hope is that decision-makers, whether at the local level, in schools, or who organizes a barbecue or wedding, will be more resilient by having the tools to make the right decisions and impose the right restrictions," Lydia Bourouiba, director of MIT's Fluid Dynamics Laboratory for disease transmission, told Business Insider shortly before the tool was published in BMJ.

"We equip people with knowledge to adapt to different situations and know when they need to be alert and when to let their guard down."

What you need to remember when assessing the risks of any situation is to take a look at where you are, look at the density of people and what they are doing.

Scientists like Morawaska say that in terms of how the virus expands, it makes no sense to distinguish a drop from the spray, or what passes to a meter from what passes to 2. What matters is how many viruses you have a chance to get into your body, no matter how you do it.

"There are three forms of contagion and all three have to be controlled," says Morawska.

Those three are people( the most common form of infection), surfaces and air.

"All this happens at the same time and making differences is difficult," explains the scientist.

Life is more dangerous in places where people are excited and excited in small enclosed spaces without good ventilation.

"Breathing, singing, coughing and sneezing generate warm, moist gas clouds of exhaled air that contain respiratory drops," said Bourouiba and the rest of the authors of the study published in BMJ.

In such conditions, even a distance of 6 or 7 meters would not be enough to be protected.

Meat processing plants are a perfect environment for contagion, because" the combination of high levels of worker contagion, poor ventilation, working conditions, background noise (leading to screams) and low use of the mask " contribute to the spread of the virus, Bourouiba and his co-authors note.

The same happens in bars, gyms, live music venues, churches and discos.

Ingenious, virologists and environmental scientists believe that you can learn to live with the virus.

"I think everyone understands what the lights of a traffic light are," Bourouiba says.

The scientist points out that it is important to have the rules in mind when considering how to reopen, meet and socialize during the pandemic.

"We need to be able to adapt and not be alert all the time," he says. "Not just with the safety distance and the mask."

The questions to ask Are: how long does the contact last? Is he wearing a mask? Is the place well ventilated? Will it be a noisy place?

"It's not as complicated to work as it takes to minimize the risk of transmission," says Morawska.

You may also find interesting: