Corporate news

The picture shows the technology using NVIDIA Jetson to achieve low-latency video transmission and intelligent analysis

2020-11-05 13:46:23 tuwei312 52


The picture shows Information Technology (Shenzhen) Co., Ltd. (hereinafter referred to as "Picture Technology") was established in 2011 and is a leading AI application solution provider. Committed to empowering businesses and individuals through AI, so that every company can use AI.



The company's main products independently developed at present are a complete set of solutions based on the NVIDIA Jetson series (edge computing), including smart cars, smart boxes in the picture, and robotic solutions. The NVIDIA Jetson platform can provide users with a set of tools for the development and deployment of AI-powered robots, drones, IVA applications, and other independent thinking autonomous machines.


The picture shows that science and technology have signed a memorandum of strategic cooperation with many well-known universities and research institutions at home and abroad, forming an integrated development of industry, university and research.


The picture shows technology

In addition, the picture shows that Technology is also a member of NVIDIA’s Startup Acceleration Program, which is a global ecological project that accelerates the development of AI startups and aims to cultivate outstanding AI startups that disrupt the industry structure.


Since the launch of the project in 2016, it currently has more than 5,000 member companies worldwide, and there are more than 700 member companies in China, distributed in more than 30 cities across the country, and covering more than 30 different industries.


The Nvidia Startup Acceleration Program provides members with market promotion, professional knowledge, and technical support through deep learning training, Inception exclusive events, GPU discounts, etc., to help AI startups grow rapidly. Scan the QR code below to join the program!


The picture shows technology


Break through the traditional network architecture, visual data processing capabilities need to be improved


At present, the traditional network architecture adopts an execution mode based on cloud computing, by deploying AI services in the cloud, relying on the rich hardware resources of the cloud server cluster to process computing requests. Although this solves the problem of insufficient hardware resources, the remote location of cloud servers causes additional delays, which makes cloud computing-based architectures unable to meet the needs of real-time services and poses considerable challenges to network bandwidth. To solve this problem, edge computing technology can be introduced to support AI services, and a large number of edge nodes can be distributed at the edge of the network to provide support for resource-constrained terminal devices to achieve edge intelligence.


In fact, the nature of edge computing is closely related to AI and the Internet of Things. At present, AI applications rely more on the cloud, while edge computing shifts intelligence from the cloud to the edge. In the future, without the support of edge computing, many applications may not be implemented. For example: autonomous driving, telemedicine, and smart cities.


The picture shows that the technology optimizes the data transmission delay to 80-120ms in the current solution, and for data structure processing, filtering a large amount of useless data, extracting accurate data, reducing the pressure of broadband transmission, and then encrypting the transmission after the structure is solved. Security issues caused by network transmission.


Jetson helps low-latency transmission of video data,

Raise the standard of intelligent analysis


The picture shows the technology focusing on the development, manufacturing, sales and service of AI edge computing solutions. At present, the smart box T100, T503, T600 and other products launched with NVIDIA Jetson series core modules have created higher standards for low latency and smart analysis. It only takes 80-120ms from data collection to processing, transmission, and application. Remote operation is like visiting the scene in person, without any delay. To realize AI, it is far from enough to solve the delay problem. The picture shows that the technology is continuously researching and improving the real-time structured processing of video data. At present, data is collected through IP Camera, GSML and other cameras, using Jetson's powerful decoding and AI processing capabilities, it can decode 1080P, 30FPS 8-32 channels, and convert the collected data into structured by algorithmic reasoning such as target detection and target tracking. The data is stored locally or transferred to a cloud server for storage. There is no need to upload all the original video data to the remote server, effectively alleviating the network bandwidth pressure. After the video data is structured and then encrypted for transmission, it also solves the security problems caused by network transmission.


The picture shows technology

The smart box based on the NVIDIA Jetson core module also has the following advantages:


The picture shows technology


The picture shows Technology CEO Su Shipeng said: “Using NVIDIA Jetson series core modules and Deepstream and Gstreamer software platforms have greatly improved visual data processing capabilities, optimized data transmission delays and structured data analysis and processing efficiency, and further improved the Internet of Things in AI Field deployment capabilities. It enables us to continuously innovate products in the field of AI edge computing and expand the intelligent application scenarios of the Internet of Things. By enhancing the AI processing capabilities of edge computing, reducing enterprise operating costs and improving production efficiency, so that more related traditional industries can enjoy early To the convenience brought by AI. The smart box products based on the core modules of the NVIDIA Jetson series can be widely used in autonomous machines such as robots, unmanned delivery vehicles, low-altitude defense, smart inspections, and smart buildings. An ideal vehicle for deep learning." 


The picture shows technology