The War In The Chinese Face Recognition Industry
摘要： “In China, there are around two hundred small-sized companies like that. However, the truth is only less than ten companies possess core technologies and four companies that can customize system integration in the country,” he revealed, stating that only a few face recognition solution providers in China while most of them are using technologies from third parties such as OpenCV and solutions from other companies that charge.
On March 14th 8:00 PM at Chongqing, 35-year-old Zhou Xi was about to start having dinner. That day he travelled from Shanghai to Chongqing in order to attend the academic conference co-hosted by Shanghai Jiao Tong University. Zhou spent five years in the University of Illinois studying his degree and later in 2011 he left NEC and came back to China to join Chinese Academy of Sciences. At present, Zhou Xi, a Ph.D holder that studies Computer Vision, has become the founder of Cloudwalk.
Headquartered in Chongqing, this startup company has already gathered over 200 talents. As AI technologies advances and gets popular, Zhou’s company subsequently gains momentum as well. “The emergence of face recognition and image identification technologies is just the beginning of realizing computer vision, which is the eyes of computers,” Zhou commented to journalists about the development of the technology, looking happy. “The ultimate goal is to build the an artificial brain, which is AI.”
Our journalists have interviewed several face recognition solution providers in China and all of them are less than 3 years old. These startup companies believe that although overseas giants such as Google and Facebook have achieved great success in the field, the turning point of the industry in China was the face recognition technology exhibition attended by Jack Ma in March 2015 in Hannover, Germany. “This very industry is young and is favored by the capital, possessing great possibilities,” This is how the world describes the Chinese face recognition technology industry.
These startup companies are based in big cities like Beijing, Shanghai, Shenzhen and Chongqing etc. and all announce that they possess some kind of original core technology. However, according to industry insiders there are only a few companies that actually have something special to offer. Statistics show that computer vision has become the most popular investment field among investors that are interested in the AI sector.
It’s important to note that giants like BAT have also set their foot in the field of face recognition. In 2014, Alibaba invested in a face recognition company and fused its technology with the big data the company possessed. In this way Alibaba pushed the development of To-C face recognition application forward. Meanwhile, Tencent has also established its own face recognition team internally. Baidu, with all the data resources at its disposal, is also developing fast.
Giants are now competing against each other in the face recognition industry. The war is about to start.
The rise of face recognition investment wave
It’s reported that Alibaba attaches great importance to face recognition technology. In the aspect of content, Alibaba has applied its face recognition technology to its platforms, including Alipay and Taobao, as well as other business sectors so as to test and nurture face recognition application scenarios.
Tencent’s Youtu team is part of the company’s social networking businesses that’s based on Tencent’s whole social networking platform, providing image support for QQ Zone, Tencent Map, Tencent Games, and other 50 products. Everyday, there are 200 million users uploading pictures to QQ Zone and the highest record of the number of processed pictures per days has been 600 million. By far the team has processed over 30 billion pictures in total.
Jack Ma erupted the trend of face recognition in China, but the attention from the giants started with the rise of AI revolution.
Some people might get the impression of what face recognition technology can do from the Movie Fast & Furious 7. The SkyEye in the movie can accurately target and locate every person on earth as long as there are camera devices. However, the truth is such technology is a integration of many, not limited to face recognition. It’s a comprehensive system that integrates recognition technologies of face, clothing, haircut, posture, walking pattern, body shape, and voice etc. Recognition technologies of face, haircut, walking pattern, and body shape are all empowered by computer vision while voice recognition is another thing.
Anyhow, companies in China have come up with excellent recognition solutions in these fields. And SkyEye represents the public’s perception on face recognition technology.
Face recognition technology proves to be appealing for many investment organizations, which contributed greatly to this current trend. Industry insiders stated that 2015 had been the year of face recognition technology and this year there would be more To-C products emerging. However, there are also people believing that in the year of 2015 and 2016 the market focuses more on the investment and layout of the face recognition technology and the turning point would be 2017-2018.
On March 15th, the founder of PCI Liu Wei stated: “Nowadays, whether it’s in China or in the global community, there hasn't been any industry standard for the fact that this industry remains in a chaotic state. But in the future, companies with true potential will rise.”
According to our journalists’ observation in recent years, Liu Wei had mentioned the importance of smart manufacturing for multiple times in public. In April 2015, PCI participated in the investment of CloudWalk, founded by Chinese Academy of Sciences Chongqing Institute. This initiative is noted with great importance in smart production.
“The application of cutting-edge technologies has always been a problem,” Liu Wei said, admitting that getting a new technology applied in real practice such as metro system, railway, and public security is very challenging. “This is what PIC is short of in some way.” CloudWalk and PIC can help each other out with their own respective strength in this sense. “We focus on computer vision and AI,” shown on CloudWalk’s official website.
Unlike Zhou XI’s CloudWalk, Face++ mainly applies face recognition technology on Internet products. Insiders told our journalist that Face++ does indenpent research and has achieved success in the application on Internet products such as Meitu and Taobao. It’s said that currently the technological level of face recognition device is far from mature. Instead, such products are very simple.
However, face recognition is after all a complicated process in technological sense. Whether it’s doing the business of To-B or To-C, every company has their own core algorithm. The situation is more like a big market where there are lots of merchants. These merchants copy the appearance and the structure of the products from giants. However, the thing is they can’t get their hands on the core algorithm, therefore they can only adapt face recognition technologies from other companies overseas or at home.
According to the source, the market at present is extremely chaotic. However, small-sized companies still mostly rely on copied devices such as tablets etc. It’s still very difficult to make dynamic identification.
“In China, there are around two hundred small-sized companies like that. However, the truth is only less than ten companies possess core technologies and four companies that can customize system integration in the country,” he revealed, stating that only a few face recognition solution providers in China while most of them are using technologies from third parties such as OpenCV and solutions from other companies that charge.
Being approached by Lenovo and Alibaba, Face++ has gathered a team of over 100 talents and had attained its A round financing at one million RMB in 2013 and B round at 22 million RMB in 2014. “In China, companies that are doing well have all got their financing,” said the source, stating that the development of AI technology is actually very expensive.
CloudWalk’s core competitiveness lies in its face recognition system and IBIS. The company’s another R&D director Li Xiafeng told our journalist:”CloudWalk connects hardware, R&D and technologies, making it a whole chain model.” Face recognition technology in most situations needs to be deeply customized according to the use so as to give quick feedback, modification, and standardized user experience to the clients.
“Currently in the market many banks’ financial certification and remote certification platforms adopt CloudWalk’s technology. To achieve this we needed more than just competitive technology, our status, which is the only company that’s been invited to draft and set the national standards of face recognition, also plays a great role. We have cutting-edge technology and strong R&D background.”
What are the reasons for such popularity?
At present, the competition in the face recognition industry has entered into a heating-up period. So when is the turning point during which companies with potential will defeat the weak ones? “When this heated popularity of this industry dies down, the turning point will eventually come. When investors burn out their money, they will tend to put money on companies with better potential.
The capital doesn't have a clear investing strategy in the face recognition strategy and doesn't have enough understanding of it,” Liu Wei told the journalist. “Before the second half of 2014 investment in the face recognition industry has been a failure. What face recognition companies should do is dig deeper into their core value and enhance the competitiveness.”
“After a few rounds of elimination, product form and operation model will grow to the point where users are able to accept. The rest will be the industry leaders.” On March 15th, Wu Deyao, director at the Bank of Xi’an’s technical department, told journalist that everybody nowadays are still competing with each other over tech specs. “A few year later the competition will be about the experience. Companies without enough experience will be out of the game, therefore they need to hurry up and participate more,” Wu added.
“Banks have many application scenarios. In the future there will be smart integration that will install face recognition technology at the counter desk, lobby and even on mobile bank clients. The truth is it’s already happening. For example, identification at the counter used to be manual work, which later evolved to be computer and machine added, further improving the user experience and shortening the operation time.
Nowadays it’s the time of high-speed photographic apparatus, but it’s moving forward to the era of face recognition technology,” Wu Deyao told journalist. “Face recognition technology will be the biggest revolution among all in all time.”
So what the capital finds in the face recognition industry? On March 16th, one of Jieao Capital’s senior executives revealed in an interview: “The application of face recognition technology has risen, which we believe is the direction with great potential. Before investing in CloudWalk, we also met with many similar companies. However we eventually settle down on this. So we together with PIC poured in 60 million RMB.”
But is this investment heat merely a fad or it’s a logical thing? On this issue, the source stated that the application and scenarios of face recognition technology has become a tendency, and that it’s up to the development speed of the market to decide the investment scene. “If there are other really good companies, we will still invest in them. We have also invested in a tech company, which is also in the technology-centric industry that will grow in the future.”
In any emerging industry, it’s very unlikely for companies to succeed fight alone. The chances are companies will fail. All the sources and industry insiders we have mentioned above all emphasized on one very important fact: the face recognition industry needs everyone’s efforts to rise higher. “If everybody is up for it, then the situation will be very really good for the industry. The strength of many overpowers that of a few.”
[The article is published and edited with authorization from the author @Bo Xiaolong, please note source and hyperlink when reproduce.]
Translated by Garrett Lee (Senior Translator at PAGE TO PAGE), working for TMTpost.