Based On Automatic Contact Tracing And Cough Detection Analysis Methods For Emergency Prevention And Control In Public Places

With the outbreak of Covid-19 in 2020,infection and death cases continue to rise worldwide.The medical industry and researchers are committed to finding new technologies and solutions to monitor and control the virus.Despite various effective prevention and control measures taken globally,the virus has not yet been fully controlled,and other infectious diseases such as influenza have never disappeared.So early diagnosis and prevention and control are still key tasks in preventing and curbing the spread of the virus.The World Health Organization has officially announced that people must maintain a safe social distance and wear masks in public places such as airports and stations.The primary task of epidemic prevention in epidemic outbreak areas is to quickly identify the source of virus transmission and conduct epidemiological investigations on contacts of infected individuals.Secondly,screening infected patients in a large population,whether they have symptoms or not,has a significant impact on the results of epidemic prevention.Especially in densely populated public places,it is of great significance to study fast and accurate contact tracing methods and Covid-19 screening technology.Therefore,this thesis focuses on the emergency prevention and control of public places in the context of the Covid-19,aiming to propose the following solutions for the prevention and control of infectious diseases:1)Research and implement an automated contact tracing method based on camera video analysis.This thesis applies technologies such as object detection and pedestrian recognition to propose a new auxiliary method in the flow survey process.Obtain close contact with pedestrians in public places within the coverage of the camera.The contact tracing technology based on Faster-RCNN and pedestrian recognition model implemented in this thesis has achieved effective results.2)Research and implement a screening method of Covid-19 based on cough detection.The traditional PCR testing method has many drawbacks and can bring many problems to the epidemic prevention system.A new type of AI voice recognition cough detection technology can quickly and accurately screen suspected patients for positive results.Users only need to upload their own cough sounds to quickly obtain diagnostic results on whether they are infected with the virus.In this thesis,the cough model combines machine learning and neural network methods,and uses transfer learning to solve the problem of small samples,and combines manual audio feature extraction method and VGGish audio feature extraction method to further improve the model performance.3)Combining automated contact tracing and cough detection methods,study epidemic prevention measures in emergency situations in public places.This thesis implements an automated contact tracing management system and cough detection program,which can effectively assist in optimizing current epidemic prevention measures.