摘要
【目的】利用深度学习的人脸检测技术为课堂教学评价提供新的方案。【方法】构建适用于课堂教学的人脸检测级联卷积神经网络模型,并进行相应优化,提出了基于统计人脸检测的抬头率来量化课堂关注度的方法。【结果】通过检测和统计课堂视频中的人脸,计算出学生的抬头率,统计出学生的课堂专注度以及时间分布,帮助教师及时准确地了解课堂教学情况。【结论】通过大量的测试和优化,该系统在人脸检测中具有较好的有效性和可靠性,可以为学生提供个性化教学,同时为教师提升课堂教学质量和教学效率提供参考。
[Purposes]Based on the technology of deep learning face detection,a new project is provided for classroom teaching evaluation.[Methods]Combined with face detection,it constructs classroom appropriate face detection optimized cascade convolutional neural mode,and proposes a new method to analyze class focus based on head-lift rate.[Findings]Through the detection and numeration of the students’faces in the video,the head-lift rate is calculated to analyze the class focus and time allocation,in order to help the teachers to be acquainted with the class teaching situation.[Conclusions]With a multitude of tests and optimization,this system possesses validity and reliability in face detection,which also offers an individualized learning for students and enhances the classroom teaching quality and efficiency for teachers.
作者
唐康
先强
李明勇
TANG Kang;XIAN Qiang;LI Mingyong(College of Computer and Information Science,Chongqing Normal University,Chongqing 401331,China)
出处
《重庆师范大学学报（自然科学版）》
CAS
北大核心
2019年第5期123-129,共7页
Journal of Chongqing Normal University:Natural Science
基金
国家自然科学基金(No.11471063)
重庆市教育委员会科技项目(No.KJ1600322
No.KJQN201900520)
重庆市自然科学基金(No.cstc2014jcyjA40011)
关键词
深度学习
人脸检测
抬头率
课堂专注度
教学评价
deep learning
face detection
head-lift rate
class focus
instructional evaluation
