Research On Personnel Identity And Information Identification Technology Based On Mobile Video

China's various industries are in a stage of rapid development,which makes the transportation hubs,factories,warehouses and other important industry places have more hidden safety hazards,and all kinds of accidents have occurred frequently.So,The key to ensuring the normal operation of various industries is to form a new highlevel rescue model.This article focuses on the emergency rescue work in various complex scenarios.For the remote transmission system of mobile video at the accident scene during rescue,for the return of video data,for the identification of related trapped persons and the intelligent identification of information,etc.,research and implementation.In emergency operation scenarios,in terms of mobile video acquisition,there are problems such as complex scenarios,difficult signal transmission links,and limited transmission bandwidth;In terms of intelligent identification of the identity and information of trapped people in the collected data,there are limitations such as limited front-end computing resources and mobile video quality deviation.This article adopts WIFI multi-level bridge emergency multimedia transmission method and system combined with mobile video-based personnel identification and information recognitionsimulates the actual video transmission environment and effectively solves the problems in information collection..In terms of information collection,this paper has formed a face detection algorithm based on improved SSD,which meets the application conditions of real-time detection in emergency rescue scenarios,and can effectively overcome the instability of video pictures in the transmission picture and the occurrence of multi-scale detection targets.The improved face retrieval algorithm based on Arcface is researched,which can achieve the best results in reducing the intra-class gap and increasing the inter-class gap in the classification process.In the rescue process,the identification process of the relevant people is as strict as possible,and the probability of misidentification is minimized.Set up an age information recognition network based on feature map overlays.Unlike traditional networks,shallow feature overlays are used as a form of deep input,while increasing the network depth,it will not increase the network calculation amount,and effectively use contextual information to extract the diverse features of faces.When face image data of different quality is taken in a mobile video captured in an emergency scene,a more accurate analysis result can be obtained.Build a data set and train each network to form a series of models and implement the corresponding functions.For identity recognition,it can achieve a comparison and authentication speed of 0.56 s,a single comparison speed of about 1ms,and an authentication accuracy of 93%.Age attribute recognition is used as a post-rescue processing work and as an auxiliary reference for rescue work,accuracy can reach 88.06%.In terms of system implementation,first,,research and develop related equipment based on 5.8GHz WIFI multi-level bridge technology.Based on wireless ad hoc network technology,improvements have been made in terms of transmission capabilities and adaptability.Can effectively adapt to long-distance,occlusion and vertical signal transmission scenarios.It is proposed to use a widely used public network 4G signal to build a communication system,combined with mobile video shooting technology and equipment.Thus the overall bridging and communication system can meet the mobile video transmission requirements in a complex environment similar to the subway environment.In order to ensure the shooting and transmission of mobile video in the emergency rescue scene.Simulated the real emergency communication environment,combined with a series of algorithm models in the information acquisition module,design and build a human identity and information recognition system based on mobile video.In the Beijing subway environment,it was proved that the relevant algorithm of the information collection system meets the application requirements in actual scenarios by verifying the various functional indicators of the system.