Research And Implementation Of Pose Face Frontalization Using Generative Adversarial Network

With the development of the intelligent age,the social requirements for face recognition technology are also deepening with the demand of related fields.In the actual scene,in order to quickly obtain someone's identity information,it leads to the acquisition of a negative face image,which may make intelligent devices confuse the identity information of the face in the image,resulting in the decline of the utilization rate of pose face image.Based on the above reasons,after researching and experimenting on the relevant methods of posture face frontal,this paper chooses to research and implement the posture face frontal method using generative confrontation network based on the comparative effect of the output positive images.In order to solve the problem that the probability of feature distribution of sample data is not balanced,which leads to the instability of the generation effect and the change of the identity between the face image and the input face image,this paper improves the theory of generative confrontation network.The main research and work are as follows.(1)The module of 3D face reconstruction is designed.The qualitative pose face image is obtained through the steps of delimiting face area,extracting feature points of face area,constructing 3D face model,texture recovery and generating pose face image.(2)By adding 3D face reconstruction module and face recognition module,the frontal face image can retain the original identity features and learn more about the potential relationship between pose face and frontal face features.At the same time,identity loss and pixel-based loss are used in the loss function to make the positive face more realistic.Finally,different face data sets are used for experiments and analysis,and the comparison results verify the positive effect of the proposed method,which effectively improves the utilization of pose face images.(3)This paper implements the frontal face system,encapsulates and applies the frontal face model in this paper,and also includes the design and implementation of user management,face area delimitation and extraction of face feature points and other functional modules,and describes and shows the use method and operation effect in detail.