BEIJING, Nov. 13, 2017 /PRNewswire/ -- At 1:00 pm on October 29, 2017, Intel "Artificial Intelligence Changes Security" Summit was held at Sheraton Hotel in Shenzhen. At the conference, Intel announced that, in conjunction with the leading AI+ solution provider ICETech, NVR and IPC based video-structured solutions will be released. ICETech deep learning video structure algorithm, based on Intel Apollolake CPU+GPU, achieved full-real time processing. Intel is commited to practicing artificial intelligence into security IoT and promoting smart security acceleration landing.
In recent years, with the rapid development of the artificial intelligence and the breakthrough in the artificial neural network optimization, the application of the artificial intelligence has expanded into many industries continuously energizing and reforming social production and people's lives. The security industry, as a set of high-definition video, intelligent analytic, integration of cloud computing technology industry, has developed from single security industry to multiple industries. It has elevated the production efficiency, improved the life to the direction of Intellectualization, and provided users with more deep learning intelligence solutions. Among them, artificial intelligence is inseparable from the big data and high-intensity computing; meanwhile, the security industry is a field with huge amount of data and complex dates, both bound to face a new wave of challenges in the ever-changing AI development.
Current situation on Deep Learning & Artificial Intelligence
Artificial intelligence cannot do without big data and high-intensity computing. Deep learning is precisely one of the core drivers and technologies based on the unique attribute of "big data + big model + large platform". Intel thinks that the full use of AI at different industries is helpful to create maximum social value. However, its study is still on the way of supervised learning. Application is concentrated on four fields: financial loans, online shopping, security and voice recognition. In the future, if Intel wants to move from technological upgrading to industrial upgrading in the field of AI. It needs to make the breakthroughs in computing capabilities, i.e. the enhancement of artificial intelligence capabilities and continuous breakthroughs in performance and adaptability.
The Challenge of Deep Learning & Artificial Intelligence
As a form of artificial intelligence applications, deep learning requires constant data training and the algorithm adjustment in actual performance which also requires the support and the storage of massive data and powerful computing capabilities, whether the video surveillance view is clear, the transmission is efficient and reliable, process relevant data are ready in time and effective feedback are provided by the back-end data center, are the primary challenges that security enterprises are likely to encounter.
For instance, there are many types of data information involved in the security industry. In terms of data structure, it includes all kinds of unstructured, structured and semi-structured information. Unstructured data mainly refers to video and image recordings, such as video surveillance, vehicle high way check point capturing pictures, face-capturing images and alarm-capturing images. Structured data include maintenance data records, analysis of structural description records and the related information database. Semi-structured data include face modeling data, fingerprint records etc. For continuous monitoring, every minute and every second is useful. The frequency of such updates is extremely fast and is occurring in large numbers almost at all times.
To quickly and efficiently filter valid information from these small amounts of information, security enterprises need to have following capabilities. Firstly, enterprises should improve the capability and the efficiency of unstructured information processing, and the ability to accurately and quickly process video images and extract useful information. It is necessary to extract useful information from the large data and the featuring modeling data, and to describe some extent of the information. Secondly, enterprises should improve big data processing technology and platform framework; provide rapid retrieval mechanism for security data information; form a targeted data information processing architecture; rapid retrieval and analysis from structured and semi-structured information.
Developing Trend of Deep Learning & Artificial Intelligence Technology
Deep learning and artificial intelligence computing represents a new wave of computing that plays an increasingly important role in resetting the core competitiveness of an enterprise. The more data an enterprise has, the more accumulated business information and the greater the value it gains. It brings an opportunity as well as higher technical requirements for the security industry.
First of all, security enterprises need to intensify studies of application in computing deeply, while they complete the security system and application. At the same time, to improve network compression and R&D abilities, the performance of hardware accelerators and the ability of processing, and to handle massive data. Especially, the most urgent subject concerning accuracy of the intelligent video applications, as well as the analytic, are the problems that need to be solved by security enterprises.
The video structure solution released by Intel & ICETech based on NVR and IPC
As the provider of AI+ solutions, ICETech has been exploring artificial intelligence for nearly 10 years. Starting from the establishment of the algorithm team in 2012, to formally beginning the deployment of deep learning technologies and related algorithm precipitation, it is said that ICETech has achieved a qualitative leap. At this time, depending on the video structure solutions created by Intel and ICETech, based on NVR and IPC, both sides aim at deepening the exploration of the advanced learning techniques and co-creating an AI platform.
AI NVR Solution
The platform integrates ICETech deep learning technology. Intel based processors, and powerful video processing capabilities of Intel ® Apollo Lake processor core graphics, enable one-of-a-kind integration of NVR, AI deep learning processing and on-wall controller and H.265 hardware codec and display capabilities. H.265 hardware decoding and display functions, and scalable hardware intelligent analysis acceleration module can be regarded as the most ideal and best cost-effective products, which can identify efficiently more than 1500 vehicles' models, brands, sub-brands, years, car body, colors and license plates and differentiate between vehicles, humans beings and non-vehicles with a high accuracy. The platform provides a powerful, stable and sensitive security monitoring system for the users.
With the rise of intelligent terminals, the enrichment of the deep learning scenarios and the convergence of the computing storage will accelerate the application of AI. Intelligent terminals and computing storage edge will become a trend. The video structure solutions created by Intel and ICETech based on NVR and IPC is a new milestone for the security industries. In the future, ICETech will continue to develop in the algorithm, constantly improving the technology. We believe that, in the near future, we will bring a new surprise to the security industry surprise.
SOURCE Beijing ICETech Sciences & Technology Company