Research And Application Of Intelligent Monitoring Technology For Online Learning

Under the leadership of "Internet + Education",especially in the epidemic era of2020-2022,online learning has been widely used in the field of education and has become one of the essential learning methods for all learners.However,at present,when learners study online,the phenomenon of machine brush class substitution is serious,and the quality and efficiency of learning are greatly affected,leading to a significant decline in the quality of education.In order to effectively monitor students’ learning status and improve learning efficiency,this project conducts research on student face recognition and real-time face monitoring in online learning to realize real-time monitoring during students’ learning process,effectively prevent the substitution and swiping situation,and improve learning efficiency.There are two mainstream facial feature recognition techniques,one is based on traditional machine learning and the other is based on deep learning.The traditional machine learning methods include face recognition methods based on principal component analysis,new SVM-based image feature extraction algorithms,etc.However,the traditional machine learning methods have the problems of insufficient feature extraction and poor model robustness.Deep learning-based methods include convolutional neural network-based face recognition methods,Deep Face-based face recognition methods,Face Net-based face recognition methods,etc.,but the existing deep learning-based methods are affected by factors such as light,pose,and background environment or do not consider the temporal characteristics of the image,which leads to the accuracy rate in the existing face recognition models The problem is still not high.To address the above problems,this paper focuses on the key technologies of face recognition and in vivo detection,and the main research contents include.(1)A convolutional neural network face recognition method based on Bi LSTM and attention mechanism is proposed to address the problem that the accuracy of face recognition is still not high due to the influence of light during image acquisition.By adding attention mechanism into the CNN network model structure and integrating feature information from different channels,the network robustness is enhanced and the ability to extract facial features is improved.Then Bi LSTM method is used to extract the time features of photos of the same person from different angles or at different times,so that the convolutional block can obtain more facial details.Finally,we use the cross-entropy loss function to optimize the model and achieve more accurate face recognition.The experimental results show that the improved network model shows better recognition performance on some public data sets,such as CASIA-Face V5,LFW,MTFL,CNBC,ORL,etc.In addition,the accuracy rate can reach 99.35%,96.46%,97.04%,97.19% and96.79%,respectively.(2)Aiming at the problems of photo fraud and the accuracy and robustness of real-time video face recognition,a real-time face detection method based on blink detection is proposed.Firstly,the texture features of image are extracted by LBP algorithm,which eliminates the problem of illumination variation to a certain extent.The extracted features are then input into the Res Net network,and the ability to extract facial features is enhanced by adding an attention mechanism.At the same time,Bi LSTM method is used to extract the time features of images from different angles or at different times to obtain more facial details.In addition,the integration of local and global features is realized through SPP pooling.Finally,the face key point detection technology is used to calculate the EAR value of the eye to realize face anti-counterfeiting,and then realize real-time face recognition anti-fraud.Experimental results show that the proposed algorithm has good accuracy on NUAA,CASIA-SURF and CASIA-FASD data sets,which can reach 99.48%,98.65% and 99.17%,respectively.(3)In this paper,an intelligent monitoring system for online learning is implemented using face recognition and live detection.Face recognition is used for fast login and automatic monitoring during the learning process(every 2 minutes the system automatically detects whether a learner is alive or not).The system can realize the function of online learning for learners and monitor the learning process in real time,which can be applied to various online learning platforms and has the value and significance of promotion.