| In recent years,face recognition technology based on human facial features has attracted much attention in the field of computer vision and has gradually become a hot research topic.Compared with the traditional manual roll call or identity card attendance,the attendance method based on face recognition has the advantages of short time consumption and high accuracy.An attendance system based on face recognition for the classroom scene is designed in this thesis.The main research work includes the following aspects:1)In order to improve the speed and accuracy of the MTCNN face detection algorithm in attendance scenarios,an MTCNN face detection algorithm based on receptive field enhancement is proposed in this thesis.First,add receptive field modules to the R-Net and O-Net networks of the MTCNN algorithm to improve the model’s distinguishability of features and the robustness of small targets.Secondly,two methods of batch normalization and global average pooling are used to improve the convergence speed of the network and avoid network overfitting.Finally,the network task weights are adjusted,and the face key point positioning tasks of the P-Net and R-Net networks are removed,reducing the training time of the model.In this thesis,the improved face detection algorithm is tested on the FDDB dataset.The experimental results show that,compared with the original algorithm,the improved MTCNN algorithm not only has faster recognition speed but also has higher recognition accuracy.2)In order to improve the accuracy of the face recognition algorithm in attendance scenarios,a lightweight CA-Mobile Face Net face recognition algorithm by combining the Mobile Face Net face recognition algorithm and the coordinated attention mechanism is proposed in this thesis.The algorithm enhances the network’s attention to image channel information and location information by using a coordinated attention mechanism,and uses Elastic Face Loss,which is more suitable for complex real-world scenes,instead of Arc Face Loss for training.Through experiments on multiple face recognition evaluation datasets,it is proved that the CA-Mobile Face Net face recognition algorithm proposed in this thesis has better recognition performance.3)Based on the aforementioned improved face detection algorithm and face recognition algorithm,a face recognition attendance system applied to classroom scenes is designed and implemented in this thesis.The system uses the Flask framework to realize front-end and back-end data interaction,uses the Bootstrap framework to design the front-end page,and uses the My SQL database to manage data.The system achieves functions such as user information management,course management,face image upload,face recognition attendance,and attendance information management.The interface of the system is simple and easy to use,and its performance is excellent,which significantly improves the efficiency of classroom attendance and saves the time of teachers and students in class. |
