Research And Implementation Of Android Social Software On Face Recognition Based On Deep Learning

With the rapid development of mobile Internet technology and the popularity of mobile intelligent devices,mobile social software receives extensive attentions.At present,the traditional identification technologies have some limitation which can't meet the needs of users,such as inconvenient to carry,easy to lose and so on.Biometrics,which has the advantages of uniqueness and portability,are increasingly and extensive used on the market.Traditional face recognition technologies can't achieve satisfied recognition performance under the huge amount of data and unrestricted conditions.Deep learning technology can effectively improve the face recognition performance under unrestricted conditions.This article introduces the current development status of mobile social software,analyzes the problems of information falsity and security existing in current social software,and puts forward to utilize face recognition technology to ensure the authenticity of the user information and the information security.In view of the fact that face recognition technology is affected by light,angle and expression,the article proposes an improved DBN face recognition algorithm.The LBP feature of the image is used as the input of the DBN for neural network training,which makes the DBN model has the robustness to illumination changes,angle changes,and gray levels variation.Meanwhile,the DBN continues to keep the characteristic of autonomic learning.To solve the problem of over fitting phenomenon caused by insufficient data set in the DBN training process,the article uses the Dropout algorithm to improve the DBN model,which can effectively prevent the over-fitting phenomenon and improve the generalization ability of the DBN model.Finally,the improved DBN face recognition algorithm proposed in this article is applied to “I do” software.In this article the whole social software based on the face recognition authentication method is developed.The experimental results show that the algorithm gets better recognition performance under the unrestricted conditions.And the Android social software has comprehensive functions,smooth operation,good adaptability and fantastic user experience,which is worth to be promoted in the market.