See how Hikvision realizes the evolution of security capabilities with AI
Jan 28, 2019

 Define AI capability implementation   At the 2017 World Summit on the Internet of Things in Wuxi, Deputy Minister of Industry and Information Technology Luo Wen said that the global Internet of Things market in 2016 is nearly 70 billion US dollars. It is expected that the market size in 2018 is expected to exceed 100 billion US dollars. What is worthy of the security industry personnel is that in the IoT information dimension, nearly 80% of the data belongs to visual data, which means that the security industry with video as the core can realize the Internet of Things industry with the extension of scene perception through video intelligence. Seamless docking, with the help of cloud technology , artificial intelligence , big data and other technologies, the future contour of smart security is becoming clearer and clearer.   However, on the other hand, although the demand for intelligent upgrade is urgent, there are not many actual implementations. A notable phenomenon is that a single technology vendor and a single hardware and software product are forming a new “technical island”. Therefore, the demand of the future industry is not only informatization, but also open, which means that only the logic of adjusting traditional security thinking can not be left behind in the era of intelligence. In addition, compared with security applications, IoT devices are complex, diverse, and widely distributed. Network, storage, computing, operation and maintenance, etc. are under tremendous pressure. Relying on cloud computing alone cannot solve all problems.   Therefore, we saw that during the AMB Expo last year, Hikvision released the cloud computing and edge computing converged computing architecture for the Internet of Things - Hikvision AI Cloud, implementing IoT + data applications and AI applications, reflecting the Internet of Things "edge The core application value of perception, on-demand aggregation, multi-layered cognition, and hierarchical application.   At this year's AMB meeting, Hikvision demonstrated the landing and thinking of more product solutions on the basis of last year, such as building a "two pools, one library and four platforms" and other products to achieve basic capabilities and data opening. Perhaps it is because the traditional thinking of the industry is still there. Some people think that it is a "fried meal," but if it is understood in depth, this is not the case. Last year, the architecture was proposed, and the corresponding software and hardware products were put into use this year. These deployments can all show that Hikvision defines the implementation of AI capabilities.   Create a "two pools, one library and four platforms"   Whether it is urban public safety or intelligent transportation , the data has less effect, the single point is weak, and the new technology has become a real pain point for users. Relying on the traditional construction mode can not really solve the above problems. In addition, the current artificial intelligence technology is still in the era of weak artificial intelligence. A large number of intelligent requirements appear in the form of fragmentation and scenario. It is difficult for one enterprise to meet all needs, and it is inevitable to construct an open AI ecosystem. trend. Therefore, it is not difficult to understand that Hikvision has been fully open in March this year after the launch of AI Cloud, and the AI industry ecosystem has been built with all parties. So what is the "two pools, one library and four platforms" highlighted by Hikvision in Beijing this year? What role does it play in Hikvision's AI Cloud ecosystem?   "Two pools" refers to computing storage resource pools and data resource pools. "One library" refers to the algorithm warehouse, and "four platforms" is the resource management scheduling platform, data resource platform, intelligent application platform, and operation and maintenance service platform. “Two Pools, One Library and Four Platforms” is a software and hardware system product that implements the Hikvision AI Cloud Fusion Computing Framework. It is suitable for the construction and upgrade of the Internet of Things intelligent in various complex environments, and solves the process of integrating the Internet of Things and the information system. Problems and challenges.   Simplified and concisely summarizing the AI strategy of Hikvision's "two pools, one pool and four platforms", that is "energy". If it is more specific, it is to realize AI landing through one-stop complete software and hardware system products .   The edge domain and cloud center in Hikvision's AI Cloud architecture are based on the “two pools, one library and four platforms” to build the corresponding capabilities, mainly to achieve the following functions:   First, the AI can be scheduled . Provide an algorithm warehouse standard framework, support third-party manufacturer algorithm access, call various algorithms and computing resources as needed.   Second, multidimensional data empowerment . Provide data management services to realize massive data link collection, cleaning association and converged storage functions such as video and RFID; provide multidimensional data fusion analysis models and mining analysis services on multi-dimensional data such as on-demand converged network data, business data and Internet data. Support smart applications and integrated applications.   Third, the application of the scene . Provide third-party application function development tools for different scenarios, support various AI+ data + video scene applications; provide basic intelligent application services such as query statistics, intelligent search, trajectory tracking, and correlation analysis; build "portraits" based on multi-dimensional data Support multi-dimensional data for early warning and control; realize "one picture" of AR real scene + data, support "three-dimensional prevention and control" application based on 3D subway.   Fourth, resource management scheduling . Provides unified management and scheduling of computing and storage resources in the edge nodes, edge domains, and cloud centers of the Hikvision AI Cloud system; compatible with unified computing storage devices, and supports cloud computing platform access management of mainstream vendors.   Fifth, the operation and maintenance integration . Provide a unified probe access framework to achieve unified monitoring and alarming of IoT resources and IT resources. Provide standardized tool management and asset management, asset management, knowledge base management, service provider management and other tools to achieve integrated operation and maintenance.   For a simple example, the traditional mode of urban management is to obtain a single information data through video acquisition (obtained as seen), but "two pools, one library and four platforms" can realize the object-aware data, user service data, Relevant data such as Internet data are aggregated as needed to obtain more multi-dimensional data on people, vehicles, places, and things, to create a more three-dimensional prevention and control system (the gain is beyond sight), through data-driven refinement of "love, refers to , diligence, supervision" four-in-one efficient service mode.   Focus on user needs   Some of the above functions and applications can be seen in other AMB booths of some of the best in the industry, and these pieces can be strung together into a relatively complete structure and applied to public security, transportation, commerce, culture and education. For many industries, to meet the individual needs of customers and have the ability to deliver quickly, I am afraid that only Hikvision will be a family.   When Hikvision AI Cloud was first proposed last year, the market turmoil caused by it was not as big as the author imagined. This year, the more "Flesh" on the basis of last year also attracted criticism, but in the era of security entering intelligent era, The concept of traditional security will gradually withdraw. In the face of the concept of innovation, the market is temporarily unable to estimate its explosive power, which is understandable.   When the new technology wave breaks out, the market will always be filled with all kinds of voices, but more vendors are still limited to single data sources, or only sell devices do not understand the business scenario, or the data real-time processing scale is low, resulting in AI solution. There are many problems with the landing of the program. Although the slogan is loud, it is difficult to create value for users and get the applause of users. On the contrary, through the new technology to meet the previously unrealized user needs, such as multi-dimensional data analysis and data association, deep mining video data value, etc., although there does not appear to be much "black technology" content of packaging, but there is enough "box office guarantee" .   The positioning of the company is closely related to the growth history of a company. In the future, whether Hikvision can inherit the style of camera technology in the network era, it still needs time to test, but in the current layout, Hikvision has the real-time processing of large scale that human beings cannot understand. The full amount of multi-dimensional data, insight into the hidden complex rules, and the ability to develop the overall strategy for the user's optimal decision-making , and Hikvision will also open this ability to share, these are enough to lead the opponent several positions. 

  • facebook