Design And Implementation Of A Cross-mirror Park Personnel Monitoring System Based On Deep Learning

With the rapid development of cloud computing,the Internet of Things and artificial intelligence,many cities across the country have successively advocated the construction of smart cities.Among them,the construction of smart parks is one of the important construction contents.However,the current park is overly dependent on security monitoring.The management of human properties leads to excessive costs and high security risks.At the same time,the management mode is passive,and management strategies cannot be adjusted for various situations in the park.At present,most of the current security monitoring is based on video surveillance,displaying the video transmitted by the camera on the display screen,manually determining the identity of the person in the video or registering the card.Some parks also provide face recognition services.They can use the gate control personnel to access and use the face information to determine whether they are people in the park.This method is only applicable to the campus access and high cost.The main problems are as follows: First,the information of personnel entering the campus or campus cannot be quickly identified.Second,it is impossible to monitor the behavior of personnel in the park.Finally,it is difficult to confirm the identity information of the person under the multi-lens information.In response to the questions raised above.This thesis is based on deep learning and combined with the latest Re ID(Pedestrian Re-Identification),also known as "cross-mirror tracking technology",which mainly solves the identification and retrieval of personnel identity across cameras.We have proposed a system for monitoring personnel in the park.This technology can be used as an important supplement to face recognition technology,and can continuously track cross-camera for pedestrians who cannot obtain clear faces.In the implementation of deep recognition-based personnel identification algorithm,we use a front-end separation mode to provide one-stop deployment service functions.According to the needs of the park,the whole system is divided into front-end display platform,personnel identification platform and campus personnel monitoring platform.The deep learning techniques used in the personnel identification process include MTCNN face detection,Arc Face face recognition and Re ID pedestrian recognition algorithm.To achieve the confirmation of the identity of the person under the multi-camera.The back-end system architecture mainly adopts the SSM development framework.The database uses My SQL to store information.In order to deal with a large amount of personnel data information,Redis is used as a cache middleware to improve the response speed to personnel search.