Faces in medical masks: China introduces a new recognition system
Chinese authorities have begun to implement a face recognition system in medical masks. AI uses key points that are located around the eyes and nose.
Chinese authorities have begun to introduce an updated facial recognition system, which can identify people in masks, scarves and even with a fake beard. They attributed the reform to an outbreak of coronavirus, due to which most of the population began to use medical masks.
In the past few weeks after the spread of coronavirus, millions of people across China have begun to wear N95 surgical masks and respirators. This led to the fact that the manufacturer of AI SenseTime adapted its product for face recognition.
The South China Morning Post notes that the technology itself is not new. In 2017, Stanford University graduate student Amarjot Singh and his team published a study on “identifying masked faces.” Their algorithm was a breakthrough in recognizing people with glasses, fake beards, scarves and hats.
“Such an AI identifies a person by identifying several key points on his face and connecting them together, forming a unique signature that is characteristic of that person,” Singh said. However, for this modification of AI, they used other key points that are open even in a masked person. The system developed by Sinh is able to recognize masked people based on key points only around the eyes and nose.
Chinese authorities have begun to introduce an updated facial recognition system, which can identify people in masks, scarves and even with a fake beard. They attributed the reform to an outbreak of coronavirus, due to which most of the population began to use medical masks.
In the past few weeks after the spread of coronavirus, millions of people across China have begun to wear N95 surgical masks and respirators. This led to the fact that the manufacturer of AI SenseTime adapted its product for face recognition.
The South China Morning Post notes that the technology itself is not new. In 2017, Stanford University graduate student Amarjot Singh and his team published a study on “identifying masked faces.” Their algorithm was a breakthrough in recognizing people with glasses, fake beards, scarves and hats.
“Such an AI identifies a person by identifying several key points on his face and connecting them together, forming a unique signature that is characteristic of that person,” Singh said. However, for this modification of AI, they used other key points that are open even in a masked person. The system developed by Sinh is able to recognize masked people based on key points only around the eyes and nose.
Photo © scmp.com
Photo © scmp.com
