Research On Crowd Density Estimation And Safety Warning Of Small And Medium Scenic Spots Based On Multi-scale Convolution Neural Network

In recent years,China's economy and society have entered a high-speed development track,and the tourism industry,as the main support of the tertiary industry,has also been developing rapidly and is facing greater challenges.Increasingly fierce market competition,forcing tourist attractions to make management model updates and upgrades,the current intelligent degree of management of tourist attractions can not match the current level of development of tourism.Especially for small and medium-sized scenic spots with relatively backward management level,how to better realize intelligent visualization management under limited cost,remote,real-time and intuitive monitoring of the main attractions and various links of the whole tourism scenic spot,realize the information management of each service scene,effectively control and avoid the orderly chaos of the scenic spot,the paper puts forward relevant algorithms for the problem of crowded safety warning in the scenic spot,and at the same time System design and information management of scenic spots for crowded scenic spots,while improving the safety of people's lives in crowded scenic spots.For the safety problem of crowded scenic spots,we firstly propose a computer vision-based perception technology,through which we can capture the image recognition of scenic spot people,and at the same time obtain the image features through the captured images,and finally carry out specific target detection based on computer vision matching;on this basis,we propose a crowd density estimation model,which is based on the pre-processing of images and foreground segmentation Based on this,the crowd density estimation model is proposed,which is based on the pre-processing and foreground segmentation of the image and the grading of this,and finally the results are analyzed for the example of Jiao Gang Lake scenic area,and the accuracy rate finally reaches 90%,which plays a high warning role for the safety of the actual scenic area in terms of human density.Through the specific characteristics and actual requirements of Jiao Gang Lake scenic area,the comprehensive platform of information management and online monitoring and security function module of Jiao Gang Lake scenic area is constructed by using MYSQL database,based on J2EE technology and AJAX page localization technology.The system is connected to the monitoring cameras of scenic spots.The integrated pictures are queried and displayed according to scenic spots information by map or table.A real-time module of scenic spots'passenger flow is designed by using computer vision technology and CSRNet-based model to realize the safety management of scenic spots' population.Through the application of scenic area intelligent management system and population density estimation model in scenic areas,to achieve the scenic area security early warning module,this study has a certain practical value.