Research On Learners’ Online Learning Status Based On Neural Network

With the maturity of multimedia technology and Internet technology,online learning is becoming more and more perfect.Online education is the product of the combination of the two.Not only that,online education also occupies two advantages of time and space,providing possibilities for different learning requirements of different groups of people,and providing solutions for the shortcomings of school education.It also enriches the current way of education.However,compared with traditional classroom teaching,online teaching cannot supervise the learning status of learners in real time,so that the learning efficiency cannot be guaranteed.Therefore,detecting the learning state of the learner is helpful to improve the network teaching system and improve the learning quality of the learner.The main work of this paper is to combine deep learning and computer vision and other technologies to study the deficiencies in online teaching,from the learner’s attention state and learning emotion,respectively.The main research contents are as follows:(1)Aiming at the difficulty of acquiring and processing real-time video,this paper proposes an algorithm for iris center and inner and outer corner of the eye based on multi-task cascaded neural network.The main idea is to use eye movement information to analyze the learner’s attention state.The algorithm can effectively obtain the coordinates of the learner’s iris center and inner and outer corners of the eyes in the real-time video,and achieves 98.9% positioning accuracy on the MPIIGaze dataset.The feature vector of eye movement is obtained from the coordinates of the key points of the eye,and the learner’s attention state is obtained by analyzing the obtained eye movement features through a recurrent neural network.(2)Aiming at the problems that traditional convolutional neural networks cannot use different levels of expression features and feature redundancy,this paper proposes a compact convolutional neural network model to classify expressions,and use the recognized expressions to learn emotions of learners.research.On the basis that the model satisfies the application of real-time system,the generalization performance of the model is improved by image preprocessing technology and optimization of loss function,and the purpose of detecting learner’s learning emotion is achieved by combining face detection algorithm.The experimental results show that the algorithm achieves an accuracy of74.71% for facial expression recognition on the Fer2013 dataset,which is superior to traditional algorithms.In order to have a comprehensive understanding of the learner’s learning state,this paper fuses two dimensions of attention state and learning emotion.The corresponding weight value is determined according to the different contributions of the two dimensions information to the decision fusion,and the final score of the learning state is calculated through the fusion information.An example analysis is carried out in the network teaching scenario,and the comparison with similar methods shows that the method in this paper can objectively and correctly reflect the learning state of the learners.