Standards are an important technical support for the high-quality development of society. The report of the 20th National Congress of the Communist Party of China clearly points out that institutional openness such as rules, regulations, management, and standards should be steadily expanded. Recently, the world's first international standard for end-to-end cloud collaboration technology, the International Standard for Digital Retina Systems (hereinafter referred to as the Standard), led by Pengcheng Laboratory, was officially released by the Institute of Standards of the International Organization for Standardization, the Association of Electrical and Electronics Engineers (IEEE SA), marking a significant breakthrough in China's standards work in this field. Wang Yaowei, one of the main research leaders of the standard and the director of the Vision Institute of the Intelligence Department of Pengcheng Laboratory, told reporters that compared to other existing industrial application standards, the standard has three major advantages: adjustable attention, and can tap the value of low-density data; The software is definable, fully optimizing the flexibility of the entire sensing infrastructure; Privacy protection effectively compensates for the lack of privacy protection in current standards. Strive to solve the problem of massive video data processing in megacities. With the expansion of urban scale and the complexity of demand in recent years, megacities have faced enormous challenges in public safety and traffic governance. Traditional visual perception systems have been unable to meet the processing and analysis needs of massive urban video data, and are difficult to support urban governance applications. "For example, traditional surveillance cameras use a 'one to one' mode, in which the transmission of massive high-definition video data places significant pressure on system bandwidth, greatly affecting the efficiency of urban governance," said Wang Yaowei. In order to solve the problem of massive video data processing in megacities, Gao Wen, director of Pengcheng Laboratory and academician of the CAE Member, took the lead in proposing the important concept of digital retina. He emphasized embedding chips in the camera to efficiently encode and compact feature extraction the visual data obtained by the camera and then transmit it to the "urban brain" for more efficient and flexible retrieval and analysis. Based on this, Pengcheng Lab established a team to conduct research and further proposed a "three stream" model based on the original "dual stream" model of "video encoding stream+feature encoding stream", which innovatively increased the model update stream, greatly enhancing the system's ability to respond to complex environments such as urban governance. Through the efficient collaboration of video streams, feature streams, and model streams, the digital retina system independently developed by the team achieves real-time processing of massive visual data and analysis and mining of spatiotemporal big data. "The digital retina standard system we proposed is based on research on the digital retina system," Wang Yaowei said. The standard system includes 12 modules, and the international standard released this time is the first module "overall architecture", fully defining the overall framework of the digital retina system. It is worth mentioning that the standard proposes a new paradigm for visual computing systems, namely "real-time feature aggregation, on-demand video retrieval, and online model update", which not only reduces the pressure on data transmission within the system, but also alleviates the pressure on centralized computing on the cloud. Moreover, the technical features of updatable models and definable functions can flexibly support various applications, Realizing real-time analysis and processing of visual big data solves resource bottlenecks in bandwidth, computing, storage, and other aspects of traditional visual systems. During research and development