Face Detection And Concentration Evaluation Method For Children’s Learning

Concentration is the ability to focus on relevant environmental stimuli.It is the basis of children’s observation,thinking,memory,imagination and other intellectual activities.It is one of the subjective factors affecting learning efficiency.With the continuous development of education,more and more people understand that most children have the characteristics of short and unstable focus duration and lack of lasting interest in activities.In the process of home school collaborative training,children’s concentration on learning at home has also attracted the common attention of parents and teachers.However,with the increase of life pressure,more and more young parents are busy with work,unable to supervise their children’s learning process at home at all times,and lack professional awareness of concentration,resulting in deviation in the school’s understanding of children’s concentration in the process of home school cooperative training.Therefore,there is an urgent need for an evaluation method of children’s learning concentration based on Deep learning technology to automatically quantify and evaluate children’s concentration in autonomous learning,so as to provide a new way for parents and schools to master children’s learning state and cultivate children’s learning concentration.This paper takes quantifying children’s learning concentration as the starting point,and takes the technologies and methods of face target detection,feature point extraction,analytic hierarchy process and comprehensive fuzzy evaluation as the main line to carry out relevant theoretical and technical research.Firstly,this paper analyzes the research status of face detection and concentration evaluation at home and abroad,and summarizes the research results of some scholars on face detection and concentration evaluation;Secondly,the related technologies and theoretical knowledge involved in the concentration evaluation are studied,which lays a theoretical foundation for the proposal of the technical scheme of the paper;Thirdly,the cascade structure of YOLOv5 s face detection algorithm and lightweight facial feature point extraction algorithm is proposed,which realizes the accurate face location and facial feature point extraction in children’s learning state,and estimates the mouth pose,eye pose and head pose;Then,combined with the estimation of mouth posture,eye posture and head posture,a quantitative evaluation model of children’s learning concentration based on Analytic Hierarchy Process and Fuzzy Comprehensive Evaluation is constructed,and the weight of each factor is calculated to realize the quantitative evaluation of children’s learning concentration;Finally,a children’s concentration evaluation system is designed and implemented to evaluate children’s learning focus online and in real time,so as to help children cultivate.The paper conducts experimental verification on the research on the assessment of children’s learning concentration.In the aspect of face detection,the average accuracy of the improved YOLOv5 s face detection algorithm in the WIDERFACE dataset is 97.8%,and the average detection time is 0.033 seconds;In terms of facial feature point detection,for children’s faces,the reconstructed and trained Mobilenet-v2 network has reached an error of 3.93% in the children’s face test set,which has reached the accuracy of the current mainstream algorithm;In terms of focus evaluation,a children’s focus detection system is designed and implemented,including posture detection and concentration evaluation module,learning log module,learning weekly report module and personal center module.It basically meets the functional needs of current users and provides a new model and platform for schools and parents to pay attention to children’s learning status.