Join Transform 2021 for the most important themes in enterprise AI & Data. Learn more.
As China’s Lunar New Year approaches, citizens and tourists will spot the country’s police force with facial recognition glasses. These glasses will help officials use real-time ID verification to fight crimes during the celebrations around this year’s event. But interestingly, such facial recognition technologies are not new to the people of China.
In the past, the country has relied heavily on cardless identification. The “Smile to Pay” feature, launched by Alibaba’s Ant Financial affiliate, lets users take a grinning selfie to authenticate a digital payment. Facial recognition in China also allows students to enter university halls, travelers to board planes, and employees to enter office premises with ease and no ID cards.
In 2015, the Chinese Ministry of Public Security announced it was looking to implement an “omnipresent, completely connected, always on and fully controllable” network using facial recognition systems and CCTV hardware. So it is not surprising that private firms and facial recognition startups in China are actively partnering with the government to monitor fraudulent and criminal behavior.
Facial recognition powerhouses
Here are some of the firms in China that have already introduced facial recognition technology to keep a watchful eye on the country’s citizens.
In 2016, Dahua Technology set a new record for labeled faces in the wild (LFW) facial recognition, beating previous records from Baidu, Tencent, and Google. Consisting of a network with over one hundred layers, Dahua’s facial recognition system is currently the deepest network among facial recognition systems. This allows a new type of metric learning method, which can detect suspects in the crowds of Beijing and even monitor pedestrians at traffic lights in the Fujian Province. Dahua says its technology helped police arrest a number of fugitives during the G20 summit.
China’s Hangzhou-based Hikvision Digital Technology claims to be the world’s largest maker of surveillance technology. The company weaves facial image modeling and a similarity calculation into the system to enable facial recognition capabilities. Users apply this solution to blacklist criminals and prevent admission of known offenders to venues like sports stadiums, parks, and casinos.
Reportedly, Hikvision facial tech has helped capture five criminal suspects picked out from a database of 973,661 facial images.
Face++’s software tracks faces using 83 different data points. The technology already powers popular applications like Alipay and Didi Chuxing. For example, the software enables Didi passengers to confirm that the person behind the wheel is a valid driver.
SenseTime has used facial recognition to help the police department nab 69 suspects in one month in southwest China. The company has a growing business relationship with the government and large data sets of a vast population, which has raised concerns about privacy. Among many other things, the company’s technology can trace the route of targets in a billion-scaled database of face image capture records and has a flexible surveillance for different times, places, and targets.
Strengths of facial recognition tech
Accurate facial recognition algorithms are based on deep learning and require a large dataset to train the system. In 2017, China built a giant facial recognition database that identifies citizens in its population of 1.3 billion within seconds, with the goal of achieving 90 percent accuracy.
The huge database and the number of partnerships the country has have helped grow China’s domination of the physical security equipment market. According to IHS Markit’s Physical Security Equipment & Services Report, the market in China will account for 38 percent of the global market by 2021. That is larger than the North American and Western European markets combined.
Limitations of facial recognition tech
Even though facial recognition allows China to fight crime and thus contributes to public safety, it comes with a long list of limitations. For one, it raises the issue of human rights. Human Rights Watch raised objections when iris scans and fingerprints were made compulsory for the residents of Xin Jiang. And some fear the government will use the information to “clamp down” on petitioners and human rights activists.
On top of this, the heavy reliance on facial recognition could lead to data hacks that leak loads of sensitive information, resulting in a security and privacy breach. Moreover, a slight inaccuracy in face detection could lead to a wrongful arrest.
Another concern about China’s reliance on facial recognition is that the technology could discriminate against people. The South China Morning Post recently reported that some restaurants employing facial recognition offered discounts based on a machine ranking of the customer’s looks. In other words, customers with “beautiful” features would get better scores and cheaper meals than those with noses the machine deemed too big or too small.
Without a doubt, facial recognition is transforming the way China does business and fights crime. However, researchers must continue to strive for 100 percent accuracy to minimize the negative effects and maximize the positive impacts.
Deena Zaidi is a Seattle-based contributor for financial websites like TheStreet, Seeking Alpha, Truthout, Economy Watch, and icrunchdata.
VentureBeatVentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact. Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:
- up-to-date information on the subjects of interest to you
- our newsletters
- gated thought-leader content and discounted access to our prized events, such as Transform
- networking features, and more