A Research On The Classification Of Students' Learning Status Based On The Detection Of Key Objects In The Field Of View

With the rapid development of the times,there are more and more temptations around students.Because some students,especially younger students,have poor self-control ability,it is difficult for students to withstand the temptations around them,and they are often affected by surrounding things when they study.This leads to a decline in learning concentration and affects learning efficiency.At this time,a method of student learning status detection is needed to evaluate the current student's learning status to assist parents and other supervising students.In response to the above requirements,this paper proposes a learning state detection method based on Faster RCNN,and combines it with judging the area of interest of students to improve the detection effect.The main work of this paper includes:(1)A method to detect the learning status of students is proposed.Take pictures of students through the camera,and use the Faster RCNN model to classify the learning status of the students in the photos.In order to improve the accuracy of classification,this paper uses dual cameras to predict the student's attention area and detect key items in the area.The system uses the Blaze Face model to detect the 2D face key points in the face area in the image,calculates the head posture information,determines the field of view of the eyes,and then uses the Faster RCNN model to detect the key items that students may pay attention to,such as books,computers,Mobile phones,etc.,thereby improving the accuracy of student learning status classification.(2)When it is detected that the student's item of interest is a computer that is turned on,this paper proposes a method to detect whether the text content on the computer is learning content.Obtain the computer page by taking a screenshot on the computer side,divide the computer screen into several parts through image segmentation,and compare several images before and after to determine the area of page change.It can be considered that the area of page change is the area where students are paying attention.This paper uses the EAST model for text recognition of pictures,the Text Rank model to obtain text summaries,and Word2Vec?Bi LSTM?Attention for text classification,which converts students' learning on the computer into a binary classification problem of text,helping to improve the accuracy of student learning status detection.The method of learning state detection proposed in this paper has an average accuracy of 93%,which is very helpful for supervising the learning of young students.