In China, facial recognition is drive for total surveillance
CHONGQING, China: For 40-year-old Mao Ya, the facial recognition camera that allows access to her apartment house is simply a useful convenience.
“If I am carrying shopping bags in both hands, I just have to look ahead and the door swings open,” she said.
“And my five-year-old daughter can just look up at the camera and get in. It’s good for kids because they often lose their keys.”
But for the police, the cameras that replaced the residents’ old entry cards serve quite a different purpose.
Now they can see who’s coming and going, and by combining artificial intelligence with a huge national bank of photos, the system in this pilot project should enable police to identify what one police report, shared with The Washington Post, called the “bad guys” who once might have slipped by.
Facial recognition is the new hot tech topic in China. Banks, airports, hotels and even public toilets are all trying to verify people’s identities by analysing their faces.
But the police and security state have been the most enthusiastic about embracing this new technology.
The pilot programme in Chongqing forms one tiny part of an ambitious plan, known as “Xue Liang,” which can be translated as “Sharp Eyes.” The intent is to connect the security cameras that already scan roads, shopping malls and transport hubs with private cameras on compounds and buildings, and integrate them into one nationwide surveillance and data-sharing platform.
It will use facial recognition and artificial intelligence to analyse and understand the mountain of incoming video evidence; to track suspects, spot suspicious behaviours and even predict crime; to coordinate the work of emergency services; and to monitor the comings and goings of the country’s 1.4 billion people, official documents and security industry reports show.
At the back end, these efforts merge with a vast database of information on every citizen, a “Police Cloud” that aims to scoop up such data as criminal and medical records, travel bookings, online purchase and even social media comments – and link it to everyone’s identity card and face.
A goal of all of these interlocking efforts: to track where people are, what they are up to, what they believe and who they associate with – and ultimately even to assign them a single “social credit” score based on whether the government and their fellow citizens consider them trustworthy.
At this housing complex in Chongqing, “90 per cent of the crime is caused by the 10 per cent of people who are not registered residents,” the police report said. “With facial recognition we can recognise strangers, analyse their entry and exit times, see who spends the night here, and how many times. We can identify suspicious people from among the population.”
The Sharp Eyes project also aims to mobilise the neighbourhood committees and snoopy residents who have long been key informers: now, state media reports, some can turn on their televisions or mobile phones to see security camera footage,
The bigger picture is to track routine movement, and after you get this information, to investigate problematic behaviour. If you know gambling takes place in a location, and someone goes there frequently, they become suspicious. — Li Xiafeng, director of research and development at Cloudwalk
and report any suspicious activity – a car without a licence plate, an argument turning violent – directly to the police.
China seeks to achieve several interlocking goals: to dominate the global artificial-intelligence industry, to apply big data to tighten its grip on every aspect of society, and to maintain surveillance of its population more effectively than ever before.
“Deep learning is poised to revolutionise the video surveillance industry,” Wang wrote in a recent report.
“Demand in China will grow quickly, providing the engine for future market growth.”
In the showrooms of three facial-recognition start-ups in Chongqing and Beijing, video feeds roll past on big screens, with faces picked out from crowds and matched to images of wanted men and women.
Street cameras automatically classify passers-by according to gender, clothes and even hair length, and software allows people to be tracked from one surveillance camera to the next, by their faces alone.
“The bigger picture is to track routine movement, and after you get this information, to investigate problematic behaviour,” said Li Xiafeng, director of research and development at Cloudwalk, a Chongqing-based firm.
“If you know gambling takes place in a location, and someone goes there frequently, they become suspicious.”
Gradually, a model of people’s behavior takes shape. “Once you identify a criminal or a suspect, then you look at their connections with other people,” he said.
“If another person has multiple connections, they also become suspicious.”
The start-ups also showcase more consumer-friendly applications of their technology. Companies like SenseTime, Megvii and Cloudwalk provide the software that powers mobile apps allowing people to alter, “beautify” or transform their faces for fun.
Much of their business also comes from banks and financial companies that are using facial recognition to check identities, at ATMs or on phones.
Facial recognition is not going away, and it promises to become a potent tool for maintaining control of Chinese society.
So far, the technology doesn’t quite match the ambition: It is not foolproof.
“There will be false positives for the foreseeable future,” said Jim Dempsey, executive director of UC Berkeley’s Centre for Law and Technology. This raises two critical questions, he said: Does a country’s due process system protect people from being falsely convicted on the basis of facialrecognition technology?
And are the false positives disproportionately skewed toward certain minority groups, such as Chinese Muslims?
“Now we are purely data driven,” said Xu Li, CEO of SenseTime. “It’s easier in China to collect sufficient training data. If we want to do new innovations, China will have advantages in data collection in a legal way.” — Washington Post/Shirley Feng contributed to this report