In recent years, with the development of high-definition video, intelligent analysis, cloud computing and big data related technologies, the security industry from the traditional passive defense to active judgment, early warning development, the industry also from a single security field to multi-industry application, improve the degree of intelligent life direction progress, for more industries and people to provide visualization, intelligent solutions. Among them, the efficiency of the intelligent video monitoring industry has contributed greatly, and the important role of intelligent video engine technology is also emerging.
As the core system of security, the core logic of intelligent video surveillance industry is: high-definition, IP, intelligent. As early as 2013, Smartcore and Huawei Hesse jointly developed the epoch-making product Hi3516A and Hi3536 processor. The realization of ultra high-definition picture quality and low bandwidth network transmission, storage and display has solved the three problems of "bandwidth, storage and display", which are recognized by the industry due to high definition and network, and brought the whole video surveillance industry into a new development period. So far, intelligent video this pair of "security eyes" in the industry has been thoroughly polished.
Along with this, how to train the flexible brain to process the "visible data" of the security eye, so that the intelligent video surveillance has become the main topic of the next development of the intelligent core. The development of "intelligent video surveillance technology" mainly includes three directions: scene adaptability, video semantic parsing and fuzzy information processing. Scene adaptability includes the ability to output high quality video analysis results in a variety of light environments. Video semantic analysis includes extracting key information features of video data and then applying them to intelligent video storage and retrieval. Fuzzy information processing includes video data analysis and processing when the front-end input video is not clear enough.
The key to support these technological progress is the improvement of the core processor performance and the optimization of related algorithms. The improvement of processor performance brings the sharp rise of power consumption and cost, and also affects the popularization of intelligent products. While meeting the needs of intelligent analysis service, high performance, low power consumption, low cost and miniaturization of core processor are the key and difficult problems for the popularization of intelligent video products. Intelligent Core original initiative "ARM+IVE intelligent acceleration engine" architecture, with extremely high performance, extremely low cost, extremely low power consumption three characteristics. This technology can harden the key software operators that consume the highest CPU resources used in intelligent video analysis, so as to improve the video analysis capability and computing speed of the whole chip, and inject new technological impetus to the video surveillance industry.
IVE is the abbreviation of English Intelligent Video Engine, that is, intelligent video engine, is the original intelligent core dynamic architecture, with the participation of a small number of main control CPU can achieve complex intelligent analysis algorithm. It can not only meet the needs of intelligent analysis business, but also greatly reduce the performance requirements of the system on the main control CPU. The system can use relatively low performance CPU, reducing the power consumption of the system, reducing the size of the system, and thus reducing the cost of the whole system, opening up a new road for the popularization of intelligent video products.
ARM+IVE: The cornerstone of Intelligent Video surveillance
ARM+IVE technology can adapt to a variety of intelligent video analysis scenarios, simple Settings, can output high-quality intelligent analysis results in a variety of lighting environments, and can extract key information features of video data for intelligent video storage and retrieval. IVE hardwareizes the complex computations that consume the most CPU resources in a variety of intelligent algorithm scenarios, leaving the weakly relevant parts of the scenario to the general-purpose CPU. In this way, the complexity of intelligent analysis implementation is reduced, while leaving the customer free play space. With the increasing number of IVE hardening operators, the CPU operation workload is reduced, and the product power consumption is reduced. With more advanced process and multi-core ARM Core product architecture, the CPU computing performance is improved and the product power consumption and cost are lower. Subsequently, customers can quickly develop and transplant intelligent algorithms based on PC environment, which is superior to DSP and FPGA development environment.
Architectural advantages of ARM+IVE:
1. Advantages of collaborative work
ARM has a lot of processing technology for multimedia;
ARM and IVE work together to maximize the advantages of both image processing.
2. Low development threshold
The development difficulty of ARM+IVE is much lower than that of DSP;
Without secret cooperation between algorithm manufacturer and camera manufacturer, video algorithm function can be developed quickly by calling IVE operator, and the development cost is small.
3. Low cost of hardware and products
ARM+IVE chips are cheaper and more efficient than other architectural solutions.
IVE as the core technology of intelligent core original motion has promoted the change of a monitoring era. From IVE1.0 only video detection algorithm, to IVE2.0 expanded license plate recognition, video detection, pattern recognition and moving target detection algorithm, and then to IVE3.0 expanded face detection and depth perception algorithm on the basis of 2.0. This "video internal analysis accelerator inside the chip" has been developing at a high speed. The remarkable characteristics of high performance and low power consumption have also been unanimously recognized by users inside and outside the industry. At present, the technology has been planned to expand to IVE4.0 version, to play a greater role in more domestic and foreign intelligent transportation, driving safety, intelligent sports, machine vision fields, let us look forward to. In the field of AI, we are working with more chip partners to create better chip-level solutions.
