If you deposit a cheque or withdraw cash from an ATM in China you’ll no longer need a debit card to do so. New facial-recognition ATMs now enable almost-instant access your bank accounts.
The tech has been deployed on over 12000 ATMs nationwide by Yitu Technology, a Chinese AI startup specialized in computer vision and machine learning. ATMs are just one facet of Yitu, whose big plan is to put smart eyes on all human-machine interfaces.
Founded by two Chinese AI experts in 2012, the company is particularly competent in recognizing human faces and vehicles. Such capability appeals to China’s public security departments. In 2015, Suzhou local police captured a burglar in just ten minutes, using Yitu’s technique to pinpoint the targeted car among hundreds in a surveillance video.
The case quickly sent a shockwave across the country. Now, over 30 provincial and 150 municipal public security departments have adopted Yitu’s technique for identifying thefts and hunting fugitives.
The China Security Industry Network (21csp.com) projects China’s security surveillance market value at CNY¥752 billion (US$120 billion) in 2018. That’s a big pie, and everyone wants a piece.
In an interview with Synced, Yitu Senior Researcher Shuang Wu said the company can accurately enable facial recognition in just one second, with a large database of over one billion facial photographs.
“Yitu’s technology has been well-accepted by many China’s crucial ports and organizations as Yitu shows a superior performance even when it comes to racial spectrum, which is usually a challenge in facial recognition,” says Wu.
Yitu’s advance in facial recognition technique quickly won global accolades as the company scored highest under the “identification accuracy” category in a competition run by the US National Institute of Standards and Technology (NIST) and the Intelligence Advanced Research Projects Activity (IARPA) in November 2017.
In recent years Yitu has been expanding the use of its vision technologies to applications in smart cities for example. Partnering with Alibaba last year, Yitu energized Hangzhou’s urban intersections by modelling vehicle behaviour and predicting traffic flow, speeding up traffic flow by 11 percent.
While Yitu has consolidated its business in the security surveillance over the years, the company has also been plotting a bold course in healthcare, a traditionally conservative field that is now increasingly driven by technology. “We believe AI’s potential in healthcare is unquestionably promising,” says Wu.
Struggling with air pollution, China has seen a sharp rise in lung cancer cases over the past 15 years. The country recorded nearly 4.3 million new cancer patients in 2015, 730,000 of whom had lung cancer, accounting for nearly 36 percent of the global total.
Yitu launched a smart medical imaging platform in early 2017 for lung cancer early detection — applying deep learning techniques to locate lung nodules, a type of small tissue mass in the lung that appear as round and white shadows on a CT scan. By analyzing size, shape and location, the system can help doctors compare a current imaging scan with historical scans to track changes in condition. It then creates a structured diagnosis report according to the imaging results, which is delivered to the doctor.
Zhejiang Provincial People’s Hospital first deployed Yitu’s system, and hospital clients have now grown to over 20. “Our stats tell us the adoption rate of our diagnosis report is about 92 percent,” says Wu.
Wu stressed that AI+healthcare is not to be treated as a short-term venture. “Public security for example has a clear threshold, and what you need to do is to reach a certain percentage of accuracy. Healthcare, however, is a bit more complicated. You have to keep talking with doctors and ask what and how to improve your product. While it is easy to make a demo, building trust in healthcare is a long-term challenge,” says Wu.
Yitu’s fast tech development and rapid rise attracted interest from venture capital. In 2017 the company raised CNY¥380 million (US$60 million) in a Series C funding round led by Hillhouse Capital Group. Last month, Chinese media reported that Yitu had raised a new round funding, bringing the company’s value to an estimated CNY¥15 billion (US$2.4 billion). The company has not officially confirmed the news yet.
Yitu now has a crew of over 500 in China, Singapore and Silicon Valley.
This yearlong series of huge funding rounds is providing Yitu with the resources it needs to develop new AI-powered products beyond machine vision, such as those based on natural language processing (NLP). Yitu recently created an NLP-based medical reference database that can efficiently convert patients’ medical records into organized and usable data.
“From a technical point of view, deep learning reinforces the inner relationship between different academic areas, or you could say the threshold for developing NLP by a computer vision company is lower. On the other hand, NLP is something that any AI company should invest in,” says Wu.
Yitu is one of four billion-dollar computer vision unicorns in China, along with CloudWalk, Face++, and SenseTime — the three-year-old startup that just became the world’s highest valued AI company. They are all vying for the vanguard in marketplaces such as robotics, finance, security surveillance, transportations, mobile device, and AR/VR.
The fierce competition between the four companies does not allow Yitu time to relax — the company must continue growing. Wu says this is an exciting time for Yitu: “There are surprises every day. A lot of things happen beyond the plan, and somehow they achieve good success, bringing great challenges as well as opportunities for everyone.”
Journalist: Tony Peng | Editor: Michael Sarazen