Skynet of China: can recognize the face of a person wearing a mask, 90% accuracy


Initially, face recognition technology was a quick and convenient way to unlock your phone. Over time, we developed this technology (and collected enough face data) to serve security purposes effectively; Many law enforcement agencies, border gates or airports have begun to apply new technology.

The question is: How good is facial recognition technology so that it can be widely applied? Especially when each face is different, and one can wear everything on the face to avoid security cameras, such as sunglasses or masks.

The above question is why Oakland and San Francisco banned the use of facial recognition technology to monitor the crowded areas; partly because they fear the results may be inaccurate, leading to disastrous mistakes. But they will have to think again soon, when many Chinese artificial intelligence companies have just announced a new tracking system that has very good facial recognition, even when the subject has a mask.

According to the research, in order to get new success, computer scientists learn to master the small parts of the face, in addition to analyzing both gait as well as recognizable body language. .

It is also common for the new technology to come from China: we already know the supernatural facial recognition capability of the Skynet system installed throughout China. It can spot a criminal among a hundred people, identify one person and know where they are moving as soon as they leave the airport, even arresting the man who has fled. Perennial.

Sitting and talking to the press, surveillance system researchers in China shared a little about the technologies behind Skynet’s success.

What will artificial intelligence contribute?

Basically, the facial recognition system will confirm a person’s identity from a database of images, including videos and photos. Using deep learning technology to speed up the face scanning process, a system will gradually learn a huge amount of data. The more data, the faster and more accurate the system.

These systems create something so called “fingerprints,” with each person having a different face pattern – just like the variety of fingerprints. It measures the distance between the eyes, the width of the nose, the depth of the eye sockets and similar facial features. With a dense network of security cameras like China, in addition to face data, other factors such as height, age, clothing color, etc. will also appear in the database.

Can masks hide the identity?

According to Chinese surveillance experts, not really.

Advanced identification systems will analyze (and predict) the details under the mask. This capability allows the surveillance system to reduce the number of matching people to low, thus making it easier to identify the person wearing the mask.

The new Chinese report is backed by scientists from abroad, in particular a team of researchers working at the University of Bradford. Last May, British scientists released a scientific report, announcing facial recognition technology that can achieve 100% accuracy with just one side of the face, half face or 3/4 face.

British researchers have made it clear that on the system only trained to recognize the entire face, the accuracy will be reduced to 40% when only the eyes and nose can be seen, and 60% when the lower half of the face is hidden. Accuracy will skyrocket to 90% when the system is trained to recognize inadequate facial recognition, achieving high results even when looking at faces without eyes and nose.

How well does the Chinese system currently detect faces hiding under a mask?

While the network of cameras and facial recognition systems are already present throughout China, their accuracy still largely depends on having to see the entire face. Many Chinese people are willing to provide personal information, including their faces, to large companies, in return for easier use of the service. Companies also leverage data from customers to help improve technology accuracy.

Just like that, the majority of Chinese artificial intelligence companies get vast amounts of data from two main sources, users voluntarily provide and data collected from their own products.

One researcher pointed out that the success recognition rate will be reduced to 70% if the user covers only one fifth of the face. However, the rate will vary based on the partial cover up.

At the moment, some Chinese AI companies are conducting system training with partially obscured faces, in order to increase the success rate in every conceivable condition.

What recent breakthroughs have enabled machines to identify even hidden faces?

The first is gait recognition technology. The new Watrix startup has introduced a software that allows a person’s gait to be detected from a distance of 50 meters, even when they have their back to the camera.

The system is trained by thousands of measurements related to a person’s gait and body structure, from swinging hands to walking. The new technology allows detection of people trying to hide their faces from surveillance cameras, masks, or just by turning their faces away.

Second, it’s voice recognition technology. It’s getting better and better from the day people still use their voices as evidence to bring them to court; Now people have used voices to activate all sorts of systems. The use of voice in combination with surveillance cameras is only a matter of sooner or later.

Finally, it’s a laser system that reads the heart rate from a distance. This is a technology developed by the United States Department of Defense, whose unique name is Jetson.

It can detect the subject’s heartbeat from a distance of 200 meters, through the infrared laser system. It can work well when the subject wears ordinary thin clothes.

The Pentagon affirms that Jetson is 95% accurate when used in the best conditions, and Jetson will be more and more effective in the future. The thing is, for Jetson to identify a person, a database of heart rate is needed first.

Can see the recognition technology increasingly advanced. Starting with the face, gradually we have both gait and voice recognition, and then even the technology seems to be only in fiction: heart rate recognition! The first three elements can be done by humans, but when we reach the limit called the “heartbeat”, it becomes clearer just how far machines are and are far from humans.

Detected and excavated dinosaur bones weighing up to 1100 pounds, very good condition in France

A paleontologist at the Paris National Museum of Natural History recently found a piece of a dinosaur’s thigh bone on July 24. This piece of thigh bone is estimated to be about 140 million years old, weighs 1100 pounds of an individual of the Sauropod plaque.

The bone, which is about 6.5 feet long, is longer than the average height of an adult, was discovered when researchers excavated at Angeac-Charente, an area commonly used to search for fossils. since 2010. This thigh bone is most certainly from Sauropod, a herbivorous dinosaur branch with very long neck and tail. This is considered one of the largest terrestrial creatures that ever existed.

Besides its enormous size, people are surprised by its very good state and integrity. According to scientists, the “owner” of this fossil could weigh up to 40-50 seals. In addition, at Angeac-Charente, scientists have been searching for more than 7500 fossils of various species since 2010.

Leave a Reply

Your email address will not be published. Required fields are marked *

Next Post

[MeetUp – August 11, 19] NeoWorld X DragonLand – Explore the New Island

Virtual world built on blockchain technology – NeoWorld will host the 2nd Meet Up with the theme NeoWorld X DragonLand – Explore the New Island. When players attend this event, they will be guided and supported directly by Vietnam NeoWorld team as well as have the opportunity to receive extremely […]

Subscribe US Now