pk10大小走势图怎么看

  1. Home
  2. artificial intelligence

AI technology industrialization is now "fault" Who will dominate the next five years?

[ Security Exhibition Network Market Analysis ] The newly released "China Semi-annual Research Report on Artificial Intelligence Software and Application Market (2019H1)" by International Data Corporation (IDC) shows that the size of China's artificial intelligence market in the first half of 2019 reached 1.76 billion US dollars (about (12.38 billion yuan), of which the banking, insurance, government, and Internet industries are still the main players in artificial intelligence technology investment.

Major changes in China's artificial intelligence market in 2019

1. Changes in market structure

The Sino-US trade friction has stimulated the hot spot in the AI market to shift from the application layer to the chip layer. At this stage, the GPU is still leading the AI acceleration market. In the next few years, NPU / XPU will be integrated into the AI infrastructure.

The market is changing from a startup company as the main body to a giant manufacturer and startup company competing on the same stage and highly competitive.

Despising the launch of IPO: It represents the success of the first batch of AI startups, and also shows the competitive pressures that AI startups face.

Traditional industry giants are making efforts in AI innovation: such as State Grid, Ping An Technology, etc.

2.New applications are incubating

RPA + AI: The merger of Laiye Technology and Orson Technology, that is, AI startups looking for potential applications in the exploration process, also stimulated the market to start paying attention to the application of intelligent process automation IPA (IPA = RPA + AI, Intelligent Processing Automation).

With video structured and intelligent process automation, digital employees will become a hotspot for exploring AI in the next five years.

3.Expanded deployment methods

The proportion of edge deployment has increased significantly, especially from end-to-side inference frameworks such as Ali's MNN, Tencent's NCNN, and Xiaomi's MACE.

From the perspective of industrial intelligence, industry + AI is mainly integrated into various processes of the enterprise in the form of intelligent application solutions. From a technical perspective, the current market situation can be analyzed from the perspectives of computer vision, speech semantics, and machine learning development platforms.

◆ Computer vision applications

In the field of computer vision, the original innovation of algorithm models has slowed down, and the focus of innovation lies in vision applications in more complex scenarios. Relatively mature application scenarios include face comparison in security scenarios, static face recognition for identity authentication, and image content auditing. The applications that have begun to move into the production environment include smart containers under new retail, product identification, product auditing, and industrial quality inspection. These applications have also created a number of new AI startups. Applications in incubation include video structuring, video analytics, autonomous driving, and more. In the overall computer vision application market, the vendors currently occupying the largest market size include Shangtang, Deaf, Yuncong, Yitu, Haidayu, and AI startups oriented to specific application scenarios such as smart eyes, deep awake technology, and innovative wisdom.

AI technology industrialization is now "fault" Who will dominate the next five years?


◆ Speech semantic application

In the field of speech semantics, algorithm innovation is still continuing. New algorithm models that will enter the market in 2019 include Bert, Xlnet, and ERINE. Judging from the application scenarios, the application scenarios occupying the largest market scale include voice assistants in the consumer market, conversational artificial intelligence customer service in the enterprise market, and spoken language assessment in the education industry and intelligence in court hearings. The following figure shows the mainstream manufacturers tracked by IDC in various application scenarios, where the area size represents the relative share of the manufacturers.

AI technology industrialization is now "fault" Who will dominate the next five years?


◆ Machine Learning Development Platform

In the machine learning market, IDC released IDC Marketscape: Chinese Machine Learning Development Platform 2019 Vendor Evaluation in August 2019, representing the market structure of machine learning development platforms in 2019H1. The current market structure in this field is mainly divided into three types of vendors: 1) Platform-level companies represented by cloud service providers mainly promote public cloud machine learning to help users quickly build machine learning models on the cloud. A large number of machine learning applications have been accumulated in the customer group. The advantage of this type of manufacturer is that it has a huge user base, which can be quickly transformed into users of machine learning products. 2) Star startups represented by the fourth paradigm and smart cubes are leading technology / product trends with cutting-edge innovative technologies such as federal migration learning and fully automated AutoML. The advantage of this type of manufacturers lies in the leading and leading technology. 3) Big data platform companies are also launching machine learning components to help users achieve predictive analysis capabilities on big data platforms, such as Xinhua III, Neusoft, and Xinghuan Technology.

AI technology industrialization is now "fault" Who will dominate the next five years?


Cloud services + AI become the main battlefield in the AI field

The main challenge of adopting AI at this stage is the lack of professional talents and high deployment thresholds. Adopting cloud AI capabilities can help users focus on the development and application of AI capabilities to a certain extent. Currently, the cloud service market has more than 100 AI capabilities. The mainstream AI cloud service vendors include Baidu, Alibaba Cloud, Tencent Cloud, Huawei Cloud, Kingsoft Cloud, AWS, Azure, etc. How to help cloud service users to intelligently upgrade and how to use AI to drive the adoption of cloud services has become the focus of the AI cloud service market.

Looking into the future, Lu Yanxia, Assistant Research Director of IDC China, said that from the perspective of the client, only 50% of the companies believe that the AI projects they have deployed have brought significant benefits, and nearly 60% of them indicate that it is very difficult to adopt AI. From the market side, the applications that can be launched and the applications that are being hatched present a certain degree of "fault". On the whole, the market pattern in the next three years will still have a lot of variables. The key to success is the ability to continue innovation, market-oriented commercialization, and the ability to help manufacturers lower the deployment threshold and accelerate landing.

This article is from the submission, does not represent my position, if reprinted, please indicate the source: /ai/55425.html