Design And Implementation Of Intelligent Video Surveillance System Based On Cloud-Edge Collaboration

Fire safety is an important part of public safety.The traditional fire monitoring system adopts the form of cloud deployment.This cloud computing model with centralized data processing as the core puts high pressure on the computing and storage load capacity of the cloud center,making it difficult to cope with the explosive growth of collected data.In this paper,combining the Internet of Things,big data,intelligent video analysis and container choreography,a cloud-edge collaboration based on fire monitoring solution is proposed to realize the cloud-edge node data collaboration,business application collaboration,service collaboration and operation collaboration.In view of the design and verification of the solution,this paper has carried out in-depth research and solution to the related problems.First of all,this paper studies the achievements of the academia and the industry in the exploration of cloud edge collaboration,summarizes the cloud edge collaboration architecture specification and the general architecture of the monitoring system.On this basis,an intelligent monitoring solution combining edge device data analysis and AI video analysis is proposed for the intelligent fire monitoring scene.This system USES a "cloud-edge" cooperative mode,the edge nodes deployment of video data pretreatment and filtering model,cloud center to perform complex computing tasks and training model,and puts forward a kind of combination of computing tasks and transmission efficiency of task scheduling algorithm that solves the traditional video monitoring of large amount of data transmission,invalid data consumes more bandwidth.Finally,based on the industry-recognized service choreography technology Kubernetes,application containerization enables unified choreography management of cloud center and edge node services,and service deployment is highly flexible,easy to expand and easy to control.The experimental results of this paper and the application effect of the actual system in the Internet of Things competition prove the availability and good portability of the system.