China’s facial recognition advances; becomes largest video surveillance network

If you cross the street in urban China these days, chances are you’re being watched. The country has built the world’s largest video surveillance network with tens of millions of cameras, and they are now increasingly capable of automatically identifying people’s basic information thanks to technological advances and big data.

Xu Chiheng, 27, a co-founder of SenseTime, a firm specializing in facial recognition, is one of the people making this happen. He is one of a generation of young entrepreneurs that’s taking advantage of advances in technology and a huge market for surveillance technology in China. Once his company completes its current round of fundraising it will be worth around $2 billion.

SenseTime and other Chinese companies have stolen a march on the competition globally. At this year’s ImageNet Large Scale Visual Recognition Challenge competition Chinese Artificial Intelligence (AI) teams swept the board, according to state media.

One of the biggest clients for companies like SenseTime is the Chinese state. The government has been actively encouraging development of the technologies, said Wang Shengjin, professor at Tsinghua University’s Department of Electronic Engineering, an expert in facial recognition technology.

While the development of such technology has had to contend with privacy concerns in other parts of the world, in China many people are ambivalent about having their every move watched.

“I think that in people’s lives there are issues of safety and issues of privacy, I think that when these two things come into conflict, Chinese people perhaps care more about safety,” said Wang.

In part because of an acceptance by the general public to surveillance, firms like Cloudwalk, who counts many local police departments as their clients, are producing technologies they say can “predict crime”.

The government is currently also in the process of building a huge database so powerful that once operational it will be able to identify any citizen within seconds through the country’s network of CCTV cameras, according to a report by the South China Morning Post in October.

Other recognition technologies aim to fill the gaps where facial recognition is currently unable to identify people.

The government is also building a database of voice patterns to be able to identify people based on the way they talk, according to a report by Human Rights Watch.

As of 2015, the police had collected 70,000 voice samples in Anhui Province, one of the areas where voice recognition trials are taking place, the report said.

Another method is through a person’s walk. Gait recognition technology is able to identity people at further distances, and doesn’t rely on a person not covering their face. This technology is being developed by Watrix, who say they have developed the most advanced commercially applicable recognition techniques in the world.

The company is currently cooperating on a small scale with local police forces to see how the technology can be implemented, the company’s CEO Huang Yongzhen, 34, said, adding that many other police forces were in contact with them.

An employee at the firm said the police force was currently testing the technology on prisoners. Like Xu, Huang deflected concerns about what his technology could do if misused.

“Before we had this technology, human skill was being relied upon to do these things, it’s just that with the technology it becomes more efficient… if you want to solve these concerns, the key doesn’t lie with the tech companies,” he said.