Face Recognition System Based On Intelligent Lightweight Glasses Convolution Neural Network

With the rapid development of wearable electronic devices,smart glasses,smart wristbands and smart helmets and other intelligent product development design has become a hot topic.These products have great application value in the fields of public safety,finance,and daily life.Through the system design and study of existing smart glasses and found deficiencies therein,and the improved method,at the same time expand the application of intelligent glasses.The main research contents of this thesis are as follows:After a review of wearable electronic devices such as glasses intelligent literature,to state of the art intelligent glasses have a more in-depth understanding,Analysis of the voice interactive technology,touch screen technology and interactive design on the smart glasses,Developed a smart glasses system that integrates voice interaction modules,touch control modules,camera,voice calls and other related functional modules.Rich aspects of the application of intelligent systems,and the technical problems encountered in the design process are summarized and analyzed and compared.Aiming at the problem that the data collection of smart glasses is not clear,a circuit design of data acquisition smart glasses based on high-speed image sensor is designed.The design and optimization of the circuit by using the camera portion of a CMOS image sensor and optimizes the data buffer circuit of the entire system.It solves the problems of poor image quality and insufficient system storage of smart glasses during data acquisition,and improves the fluency of the system.And the glasses were tested to verify the rationality and effectiveness of the circuit design.For implementing face recognition system in the smart glasses,limited hardware conditions and slow processing speed and other problems,face recognition method based lightweight convolution neural network.Enhanced data sample data set by using cropping,rotation or the like.Then using the sample data based on feature extraction MobileNet lightweight convolutional neural network,the target detector and the SSD sample data human face recognition.Finally,compare and traditional convolution neural network recognition algorithm to verify the superiority and effectiveness of the algorithm.