Research On Person's Gender Rapid Detection Method In Security Video

In the context of big data and artificial intelligence,more and more biometric recognition technologies have been applied in different fields.Video-based gender rapid detection technology is one of the research hotspots.The security video surveillance system can use this technology to quickly identify and classify the gender of a certain area.Its application areas include the maintenance of public order in the entrances of gender-sensitive areas such as public toilets,mother and child rooms,bathrooms,and changing rooms;sex statistics of store customers;statistics analysis of gender-related person behavior videos and psychological indicators.This paper designs a rapid detection method for human gender based on the real-time requirements of security video analysis.The detection method is mainly divided into three parts: face detection,face feature extraction,and gender classification.In the test platform,the average processing speed is 62 ms,which effectively solves the problem of rapid gender analysis in the video.The main research content of this article is as follows:1.Face Detection in Complex Backgrounds It mainly elaborates the principle and application of face detection algorithm under complex background,and determines the face detection method b ased on Adaboost algorithm in combination with performance requirements and application environment.This method uses cascaded classifiers and Haar features to implement face detection in complex environments.2.Feature extraction method for gender identification Mainly introduce the main feature extraction methods,such as Haar features,Directional Gradient Histogram(HOG)features,Local Binary Model(LBP),Convolutional Neural Network,and analyze their advantages and disadvantages.The convolutional neural network algorithm with more robustness and accuracy is selected as the feature extraction algorithm for gender detection in this paper.3.Research and implementation of human gender recognition algorithm Firstly,the origin of the neural network and the operating mechanism of the neural unit are described,and the production of various face data sets and data sets is introduced.By applying a streamlined convolutional neural network design,the network uses smaller-sized input data and a small number of weight parameters,so that feature extraction and network iteration time-consuming are effectively controlled.4.Test and analysis of gender recognition algorithm based on face image Mainly analyzes the reasons for the poor timeliness of traditional d etection methods,and proposes corresponding optimization schemes: data image preprocessing and convoluted convolutional neural networks.Through experimental verification,the average processing speed of this method on the test platform is 62 ms,which can effectively solve the problem of rapid gender analysis in the video.At the end of the paper,the innovation and achievements of the project were introduced.At the same time,the problems of the research and the areas to be optimized were analyzed.