基于瞳孔追踪的网络课堂注意力检测  

Attention detection in online classroom based on pupil tracking

作　　者：邓景云 赵丁皓 秦慧伶 胡锦宗 钱程 陈雅茜[1] DENG Jingyun;ZHAO Dinghao;QIN Huiling;HU Jinzong;QIAN Cheng;CHEN Yaxi(School of Computer and Artificial Intelligence,Southwest Minzu University,Chengdu 610041,China)

机构地区：[1]西南民族大学计算机与人工智能学院,四川成都610041

出　　处：《西南民族大学学报(自然科学版)》2025年第2期206-215,共10页Journal of Southwest Minzu University(Natural Science Edition)

基　　金：西南民族大学中央高校优秀学生培养工程项目(2023NYXXS041,ZYN2024116);西南民族大学横向项目(横20240096)。

摘　　要：近年来,基于瞳孔识别的注意力检测技术因其无感感知和低成本优势成为了智慧教育领域的创新解决方案.然而,由于眼睛特征的个体差异、不断变化的光照条件、多样的头部姿势以及眼动方向的复杂性等因素,现有追踪算法在实时性和精确度方面还有很大提升空间.同时,如何有效地从视线追踪数据中推断出注意力仍是一个待解决的关键问题.因此,提出了基于瞳孔追踪的注意力检测方法,首先对瞳孔数据集进行预处理,采用Adaboost算法和OTSU动态阈值获取瞳孔中心区域位置特征并结合分类和回归损失组合模型实现实时视线追踪,然后基于分割网络和决策网络模型,利用眼睛开合度和注视离散度来综合判断注意力集中程度.实验结果显示,所提方法在实时视线追踪的精度和处理速度方面均优于现有算法,且在眼睛开合度预测方面达到了96.72%的高准确率.In recent years,attention detection technology based on pupil recognition has emerged as an innovative solution in the field of smart education,owing to its advantages of unobtrusive perception and low cost.However,due to factors such as individual differences in eye features,constantly changing lighting conditions,diverse head postures,and the complexity of eye movement directions,there is still significant room for improvement in the real-time performance and accuracy of existing tracking algorithms.Meanwhile,effectively inferring attention from gaze tracking data remains a key issue that needs to be addressed.Therefore,a pupil-tracking-based attention detection method is proposed.Firstly,the pupil dataset is preprocessed,and the Adaboost algorithm and OTSU dynamic threshold are used to obtain the positional features of the pupil center region.A combined classification and regression loss model is then employed to achieve real-time gaze tracking.Subsequently,based on segmentation and decision network models,eye opening and closing degrees,as well as gaze dispersion,are utilized to comprehensively assess the level of attention concentration.Experimental results show that the proposed method outperforms existing algorithms in terms of both the accuracy and processing speed of real-time gaze tracking,and achieves a high accuracy rate of 96.72% in predicting eye opening and closing degrees.

关 键 词：瞳孔追踪 注意力检测 深度学习 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]