The Design And Implementation Of Children Picking Management System Based On Face Recognition

At present,the personal safety protection of young children in kindergartens is highly concerned by the society.The lack of self-protection awareness and judgment ability makes them vulnerable to harm when facing dangerous events.Manual recording and pick-up card are the traditional ways of picking up children.However,these methods have hidden safety hazards such as children being missed,picked up by mistake,or even being kidnapped.With the development of face recognition technology,consider combining the business requirements of kindergarten picking up management with face detection,live detection,face recognition and other technologies in the field of artificial intelligence to design and implement the child picking up management system based on face recognition.The main work of the thesis includes:(1)Based on the development status and characteristics of the kindergarten children picking up mode in China,conduct a demand analysis in accordance with the software engineering design and implementation process to determine the functional architecture of the system.Design the overall architecture of the system.The back-end server of the system is developed using the Spring framework to achieve the system's loose coupling between system modules and concurrent request processing.Use My Batis to achieve data persistence,My SQL database to store users' data,Redis to achieve data caching.(2)Design five main functional modules including children attendance,children picking up,information management,home school mailbox,and background management.Children attendance module realizes recording children attendance at school,recognizing children in the video frame collected by the camera equipment at the entrance of the kindergarten,generating child arrival information.Children picking up module realizes identifying parents who pick up children when children leaves the school,recognizing the parents in the video frame,and generating child leave information.On this basis,implement the system business process and database schema design.(3)Combining the computing power of the device and the actual application scenarios,design a light system algorithm model suitable for the face recognition scene at the entrance of the kindergarten.Design and implement the face detection model based on Mobile Net network structure,design and implement the live detection model based on Feather Net,and design and implement the face recognition model based on Arc Face and Res Net50.On the basis of the above research,implementing the face recognition model for children pick-up management.On the public data set,a multi-model comparison experiment is carried out on the three models of face detection,live detection and face recognition.The experiment showed that the parameters of the model that we design and implement are smaller and the accuracy is higher;Testing the model's actual usability on the private data set collected by the camera.The test results show that the face recognition model which is designed and implemented in this paper can better adapt to the child attendance and pick-up scenarios,both the accurate and the recognition speed meet the performance requirements.Finally,implementing the children pick-up management system based on face recognition,and completing unit testing,functional testing,and non-functional testing.The test results show that all functional test cases passed,and all non-functional indicators met the design requirements,verifying the effectiveness of the system design.