Design And Implementation Of Face Detection And Attribute Recognition System Based On Neural Network

With the rapid development of economy and technology,face detection and attribute recognition algorithms are widely used in finance,education,security,e-government and other scenarios,creating benefits for many enterprises.Detection and attribute recognition,as two core research directions of face algorithm,are paid attention to again due to the hot development situation of artificial intelligence.As a national development strategy,artificial intelligence technology has been written into the blueprint of the country's future development.Therefore,as a more easily realized face algorithm,it has become the field that many scholars and entrepreneurs strive to develop.However,considering the complex scene,real-time and accuracy requirements,the current face detection and attribute recognition algorithm is still worth in-depth study.This paper focuses on the neural network based face detection and attribute recognition technology.Firstly,the basic principles of face detection and attribute recognition algorithm are described.The idea of convolutional neural network is introduced.This paper introduces three face detection technologies: MTCNN,SSH and SFD.The key technologies of front and rear end implementation are introduced.Secondly,the system design and implementation are completed.First,set data labeling standards,collect data and evaluate the data with clarity algorithm,train the model and save the model.Secondly,tornado back-end framework is used to develop face management system and put the trained model in the background for invocation.Using opencv and deep learning framework to implement face algorithm,and encapsulated into a class,provide API for the front end.At the same time,mongodb database is used to manage data,such as account information,etc.,and document database is easy to use.Third,WeChat small program face-cueb is developed to interact with users and provide face detection,face attribute recognition,face makeup,face key points and other functions.Finally,the test of face detection and attribute recognition system can run normally,and proved that it performs well in real-time and accuracy.