At the end of the Eastern Han Dynasty, the local separatist forces grew up, and the separatist forces fought and merged constantly. The three separatist regimes represented by Cao Cao, Liu Bei, and Sun Quan eventually remained. Since then, the three forces cooperated and competed with each other, forming a pattern of Three Kingdoms.
Now in the 21st century, with the continuous development of artificial intelligence technology, it seems that we can see the situation of “three parts of the world” again, and it is in the most popular field of computer vision face recognition that this scene occurs.
Computer vision face recognition technology has entered the stage of large-scale commercial landing after early research and practice. In the early stage of face recognition technology research and development, only a few enterprises participated in it. When the technology development became more mature and the application prospect became more obvious, many enterprise players entered the game one after another. Through their own development advantages, they continued to attack the city and seize the first landing scene of face recognition business.
From the perspective of the middle and lower reaches of the current face recognition industry chain and the touch level of end users, face recognition enterprises can be divided into three camps. They are the technology-based camp of start-up enterprises with core recognition algorithm as the main advantage; the ecological camp of Internet giant enterprises with huge data accumulation resources and capital research and development advantages; and the manufacturing camp of traditional enterprises represented by hardware equipment manufacturing.
Technology based: the rise of start ups
According to the latest “2019 Market Research Report on computer vision face recognition” released by Eurovision think tank, the total financing amount of the top 10 start-ups of computer vision face recognition is about 35 billion yuan, of which the financing amount of the top 3 enterprises is more than 30 billion yuan, which is amazing.
Nowadays, the financing of AI start-ups depends not only on the touching entrepreneurial “story”, but also on the technological strength with real development prospects. Without the foresight of technological development and the feasibility of commercial landing, everything is empty talk.
From the perspective of top 10 financing enterprises, Hanbai technology was first established in 2009, the latest cloud Cong technology was established in 2015, and top 3 enterprises Shangtang, Kuangshi and Yitu were established in 2014, 2011 and 2012 respectively. All of the above are typical early start-ups of face recognition. Relying on the R & D investment in recognition algorithm, each company has obvious technical advantages and industry leadership in detection rate, false detection rate, missed detection rate and detection speed.
Industry leading technology advantage is the biggest characteristic of start-ups, but if only a single technology leading, but no landing application of hardware products as a development carrier, it may also become a fatal defect of technology-based enterprises. Just like a general who is familiar with the art of war and is good at arranging troops, if there is a lack of soldiers who are in charge, he will inevitably fall into an embarrassing situation.
The value link of the whole industrial chain is relatively weak for the technology enterprises that rely solely on software. If they can’t get through the downstream applications through hardware products, they can’t realize the integration and penetration of multiple scenarios, and their competitive strength will be obviously limited.
Therefore, technology-based enterprises usually choose to cooperate with hardware equipment manufacturers and system integrators to enable product application and lay out diversified products, so as to extend the industrial value and realize the touch of end users. At the same time, some enterprises choose to rely on their own strength to develop hardware and software integrated equipment, based on the realization of the whole industrial chain, such as Kuangshi technology and yituke Technology, etc.
Ecotype: penetration of Internet enterprises
In the commercial application of face recognition, it needs not only accurate algorithm technology, but also huge user data to support.
The Internet giants represented by Baidu, Alibaba and Tencent have formed their own strong enterprise ecology and accumulated rich user data after years of industrial cultivation.
In this context, through the development and application of face recognition technology, we can not only further enrich the product service ecology of Internet enterprises, but also rely on the mature ecological field and huge data resource pool of enterprises to quickly realize scene penetration and application landing.
In 2015, Alibaba released the Alipay face recognition technology “Smile to Pay”, which was developed by the ant Kim suit and Face++. In June 2017, Ali cloud officially released the two visual intelligent services of “image recognition” and “face recognition”.
In 2012, baidu launched the first full network face search engine in China based on face recognition technology; in 2013, baidu media cloud officially launched face recognition service; in 2016, baidu brain including face recognition technology was officially released.
In 2012, Tencent Youtu was established, and from 2012 to 2013, the face recognition technology based on Tencent Youtu was applied in QQ space; in 2015, Tencent Youtu open platform v1.1 was newly released, launching Youtu self-developed uface deep face recognition system; in 2016, Tencent Youtu team was independently upgraded to a company level Youtu Laboratory, and released a new 3.0 open platform; in 2018, Tencent Youtu team launched a new 3.0 open platform Xunyoutu laboratory has been upgraded to Tencent computer vision research and development center again, increasing R & D investment.
From the development path of Internet enterprise identification technology, it can be seen that on the one hand, Internet enterprises strengthen cooperation in algorithm technology with technology-based enterprises; on the other hand, with their strong financial strength, they continuously strengthen R & D investment to achieve controllable core technology innovation.
Relying on the rich Internet user ecology, Internet enterprises rapidly develop face recognition business externally based on their own public cloud platform, and expand horizontally through internal diversified Internet services, such as the application extension of face recognition in Alibaba financial payment ecology and Tencent social entertainment ecology.
However, due to the lack of mature national sales channels, as well as the lack of large-scale project contracting capacity and implementation capacity, the traditional Internet development mode makes the Internet enterprises in the to B and to g business field is particularly difficult. Therefore, Internet companies usually choose to cooperate with system integrators to make up for the deficiencies in the ecological development of face recognition.
Manufacturing: the advancement of traditional enterprises
The traditional equipment manufacturing enterprises are huge in size, with their own characteristics and advantages in the field of hardware products. They have been deeply cultivated in the industry for many years, and have the ability of project integration and implementation.
According to the latest “2019 computer vision face recognition Market Research Report” released by eurothink, among the listed enterprises of traditional hardware equipment, there are 28 enterprises mainly involved in face recognition technology. Eurothink sorted out the top 10 related enterprises in 2018 according to the middle and lower reaches of the industrial chain.
In terms of capability, 10 enterprises can provide hardware products, comprehensive solutions and project implementation capabilities. Relevant representative products include video surveillance equipment, financial self-service equipment, gate travel equipment, etc., covering the main commercial landing scenes of face recognition, such as security, finance, transportation, smart Park, etc.
Therefore, the traditional equipment manufacturing enterprises can easily realize commercial application only through the development and implantation of face recognition software algorithm. Although most of the algorithms used by traditional equipment manufacturing enterprises are obtained through cooperation with start-up technology enterprises, joint research and development, and authorization, more and more traditional manufacturing enterprises are investing more and more in the research and development of core face recognition technology, striving to build hardware and software integrated recognition equipment based on their own algorithm ability.
From the perspective of technological advantages, although it is difficult for traditional manufacturing enterprises to surpass start-ups and Internet Ecological Enterprises in the short term, with the investment in R & D and the practice of application, the identification technology of traditional manufacturing enterprises will be improved and improved rapidly.
On the other hand, in terms of products, traditional enterprises’ hardware product perfection, manufacturing capacity and nationwide sales network channels are difficult for the other two camps to surpass in the short term or even in the medium and long term.
Taking the channel construction as an example, the national sales network of traditional enterprises covering their respective fields not only needs a lot of funds to pave the way for construction, but also is the enterprise’s accumulated income after 10 or even 20 years of operation, with perfect comprehensive abilities of sales, implementation and service.
Therefore, by catching up with face recognition technology, traditional manufacturing enterprises can rely on their own production and implementation capacity to quickly achieve business landing in their fields, seize the share of emerging markets, and realize enterprise transformation and upgrading.
The enterprises of the three camps have their own characteristics, and they cross each other, with both cooperation and competition. In the long run, although some traditional manufacturing enterprises cooperate with the other two camps and rely on each other, when the face recognition technology is perfect and mature and there is no substantial difference in application, the traditional enterprises are likely to become the biggest potential threat and realize overtaking on the curve.
In this emerging market, the three camps of enterprises continue to make efforts to improve the industrial ecology, and create a tripartite market pattern with commercial innovation. However, whether the pattern of “three parts of the world” will be broken due to technological breakthroughs or the rise of enterprises remains to be seen.