Abstract: In the context of the strategy of building a powerful country through science and technology.the demand for intelligent video surveillance has soared in many fields, However, traditional video sur-veillance and centralized video cloud computing architectures face numerous limitations, such as insuffi-cient end-side intelligence, transmission delays, issues related to network capacity, and low computingefficiency. This paper deeply analyzes the development process and current status of intelligent video sur-veillance technology, and innovatively proposes the IAPVS (Intelligent Analysis Platform for Video Sur-veillance) design mode based on distributed edge computing. This mode consists of two core components:video edge computing nodes and a data center. Edge nodes are deployed on-site or in near-field environ-ment and the YOLO algorithm is used for video data preprocessing-accurately identify targets, filtering da-ta, and then uploading the processed data. The data center integrates muliple platforms and a model re-pository, enabling operations such as model training, correction and reuse. The logic design and workflowof IAPVS Platform are also elaborated in detail. This design breaks through the limitations of traditionalarchitectures, realizes end-edge-network-cloud collaboration, establishes an efficient intelligent supplysystem, and improves data processing and resource utilization. It provides a new overall solution for thedevelopment of video intelligent analysis platforms in diverse scenarios, facilitating the development of in-telligent video surveillance technology in China.

Key words: intelligent video surveillance, distributed edge computing, IAPVS (Intelligent analysis plat-form for video surveillance) design model, terminal-edge-network-cloud collaboration, YOLO algorithm