基于卷积神经网络的课堂人脸打卡算法

Classroom Face Register Algorithm Based on Convolutional Neural Network

在现行的学校教育中,课堂效率尤为重要,为了节约教师时间,提高课堂效率与时间利用率,提出利用计算机视觉检测课堂中的人脸,并对人脸进行识别和记录,将结果反馈给课堂教师,使其对提高课堂效率,课堂时间使用率有着重大的意义。提出使用基于Dlib的人脸检测器,搭配基于卷积神经网络的人脸识别算法来进行实现对课堂的人脸打卡。实验表明:人脸检测算法在CelebFaces图像库中有着96.7%的准确率,人脸识别算法在课堂场景下的图像中有着88.8%的准确率。

In the current school education,classroom efficiency is particularly important.In order to save teachers time and improve classroom effi ciency and time utilization,proposes to use computer vision to detect faces in the classroom,identify and record faces,and feedback the re sults.Classroom teachers make it important to improve classroom efficiency and classroom time usage.Proposes the use of Dlib-based face detector,combined with convolutional neural network-based face recognition algorithm to achieve face register in the classroom.Experi ments show that the face detection algorithm has 96.7%accuracy in the CelebFaces image library,and the face recognition algorithm has an accuracy of 88.8%in the image in the classroom scene.