Design And Development Of Mask Occlusion Face Recognition System Based On Deep Learning

Since the outbreak of the epidemic,in order to avoid the spread of the virus,wearing masks has become people’s daily habits.In order to meet the contactless living environment,face recognition technology is used to verify people’s identity in schools,enterprises,hospitals,stations and express delivery stations.Therefore,face recognition technology is facing more and more challenges,and traditional face recognition can not meet the current needs.In order to solve the problem of face recognition caused by wearing a mask,this paper will focus on how to complete the task of repairing the face of the person to be recognized and then recognizing it under the condition of mask occlusion.This paper involves the face repair algorithm,multimodal fusion framework and attention mechanism face recognition algorithm of nested generation of confrontation network,and designs and develops a face recognition system for mask occlusion based on the above contents.To reduce the impact of mask occlusion on face recognition by optimizing the quality of repairing missing images and improving recognition accuracy,the specific work is as follows:1.First,the face recognition algorithm based on the nested generation of confrontation network is studied.The algorithm mainly includes two steps.The first step is to complete the generation of fuzzy image according to the extracted feature information,and the second step is to extract the detail features on the basis of the first step to generate a clear image to complete the RGB image closer to the original complete image.It lays a foundation for face recognition.Compared with other face repair algorithms,this paper proposes a nested generation of face repair algorithm against the network.The generated image is clearer and can retain most features of the unobstructed part.2.Secondly,this paper mentioned the multimodal feature fusion framework,and analyzed the impact of the fusion level on the feature extraction effect.The earlier the fusion effect,the better.3.Combined with multimodal fusion technology,this paper proposes multimodal channel fusion attention mechanism and feature fusion face recognition algorithm based on global attention mechanism,which can better extract the features in the face image,classify the acquired features into key and non-key features,improve the effective utilization rate of features,reduce feature missing,and improve the recognition accuracy.4.Based on the above research,this paper designs and develops a face recognition system with masks.According to the experimental results,this system can complete the face recognition with masks,and performs well in the repair performance and recognition rate.