When almost everyone is still immersed in the cold winter of 2018, technology companies in the AI field have taken the lead in competition in the new year. Then, in the year of 2019, can mass production and scale application of artificial intelligence be realized? And AI chip seems to be a topic that technology providers in this area cannot avoid.
Nowadays, AI Vision is most popular in the filed of deep learning and has fully obtained the goals more likely to be implemented in computer vision. Whether in images or in videos, there are a lot of frameworks and libraries which make computer vision tasks a breeze.
In order to realize the strategic implementation of the multi-modal AI chip, in addition to voice technology, WIMI AR Hologram has made great progress in AI vision. WIMI AR Hologram has created the third-generation holographic AI technology product after years of technical research and development. The simulation degree over 98% takes the users’ breath away in astonishment. With multi-modal technology based on face information analysis, functions such as face/object recognition, expression analysis, labeling, and lip-movement tracking can be realized, which can provide more feasible and flexible product interaction and user experience.
Currently, WIMI AR Hologram acts as a developer, service provider and operator of holographic AI cloud mobile software, and has become one of the largest integrated platform of holographic AI field in the country. WIMI AR Hologram has been well developed in terms of technology. It is qualified with more than 4,325 holographic content IP reserves and owns 295 AR-related patents and 76 software works.
It is necessary to combine chip with the algorithm. In this process, the boundary is blurred. With the continuous evolution of technology, the increasing depth into scene applications puts forward more needs. At this stage, a key scenario is selected for the Internet of Things, and besides, it is asked to accomplish better combination of computing power with application services.
WIMI AR Hologram’s commercial application scenarios are mainly concentrated in five major fields of home entertainment, light field cinema, performance system, commercial launch system and advertising display system. For any patent within international patent protection field, WIMI AR Hologram will provide protection by strict patent protection and the optimal holographic AI simulation experience, ensuring the remarkable user experience in Greater China.
By the year of 2017, in the fields of short video on mobile Internet, games, applications, photography, and advertising, WIMI AR Hologram Platform has completed intelligent implantation of Augmented Reality (AR) holography for 30 million times, and displayed AR totally for more than 3 billion times. Moreover, WIMI AR Hologram has made smart portraits for 300 million users through the AI plug-in, and provided tabbed AR holographic implantation of applications according to the AI algorithms. Then, WIMI AR Hologram provides AI augmented reality insertion services for thousands of business customers, involving the intelligent AR holographic synthesis of more than 4,000 business scenarios and 600,000 applications. WIMI AR Hologram is integrated into augmented reality through high-quality IP, with more than 700 million users totally. WIMI AR Hologram takes the advantage of its subsidiary to provide AI augmented reality services for hot Internet TV series and long videos, with the audience including: 1.06 billion views on Wasd Beyond Light Years, 1.31 billion views on HUAJIANGHU, 340 million views on The Great Adventurer, and 1.28 billion views on Yi Guan Xiao Zhuan. Additionally, WIMI AR Hologram has provided more than 8 billion services of AR 3D implantation of AI scenes, characters and props.
Until now, the artificial intelligence industry has not established a large-scale market, and no company can achieve absolute market monopoly. New companies are emerging in algorithms, frameworks, chips, modules, accelerators and other fields.
Now, it is exactly the era of deep learning, when the technology is improved very fast. An increase of one percentage per year was amazing before, while there is an increase of dozens of percentage per year now. So after years of development, there have been a large number of products on the market, suggesting the huge investment by big companies. Microsoft, Google, and Facebook, all of which have been engaged in this field early, have owned their respective patents.
For example, SenseTime determines its technology patents as the core strategy. In just a few years, dozens of papers on deep learning applied to the field of computer vision and on solutions to visual recognition have been published at home and abroad. Globally, SenseTime have announced its cooperation with Qualcomm on the global artificial intelligence industry, namely Qualcomm’s chip coupled with the algorithm of SenseTime. And Qualcomm also announced SenseTime as its only partner in the field of AI. The reason why SenseTime is acknowledged by Qualcomm is the patents of SenseTime, and such patents are also the main support for the long-term development of SenseTime.
As a representative of the global artificial intelligence field, 10 papers from Megvii have been accepted by the European Conference on Computer Vision (ECCV 2018) held in Munich, Germany. From the perspective of content, these papers cover many aspects of CV technology, including new representations, new model designs, neural network design principles, determination of new tasks & new methods, and even new exploration of learning under weak supervision. Fundamentally, technological innovation lies in talents, and therefore Megvii has been vigorously cultivating Chinese artificial intelligence professionals so as to establish a high-level innovation team. It is reported that Megvii has established its advantages in world-class intellectual properties and stuck to its own talent training mechanism, and Megvii also has more than 700 patented core algorithms and technical applications totally at home and abroad, ranking first in China and taking the leading position in the world in the same field.
The combination of 5G and artificial intelligence will really promote the application and implementation of AIoT. The end side and center side require quicker response and recognition so as to achieve more complex functions. Such challenges as the large-scale increase in the number of IoT devices, the lower cost of connection, the complex and variable data dimensions, and the more vertical application scenarios will further be posed for the design of IoT AI chips.
The past few years are the best time for AI enthusiasts and machine learning professionals. Because these technologies have grown into mainstream technologies and are affecting the lives of millions of people. Technology companies in this field, such as WIMI AR Hologram, Megvii, and SenseTime, are making efforts to stand out in this competition. The same is true for data science practitioners. They have to rapidly improve themselves to keep up with the ever-changing times.