Classroom rapid roll-call and sign-in method adopting face recognition technology based on deep learning

Abstract

The invention discloses a classroom rapid roll-call and sign-in method adopting a face recognition technology based on deep learning, and relates to the technical field of computer vision, in particular to the face recognition technology. The classroom rapid roll-call and sign-in method comprises the steps that a camera at a mobile terminal quickly collects target face videos and generates an IP address; a server is connected with the IP address provided by the mobile terminal under an ubuntu system and generates pictures frame by frame; key points of the face pictures are detected according to an MTCNN algorithm, and a Procrustes analysis method is selected to conduct face aligning; according to the aligned face pictures, a FaceNet algorithm is selected to conduct face vector characterization; and collected face vectors are compared with characteristic vectors of the face pictures in a database, the face picture most similar to a target face picture is found out according to a certainthreshold, and a text file is generated. According to the classroom rapid roll-call and sign-in method, face images are quickly scanned based on a deep learning method, and the face pictures affectedby the factors such as makeup, light and the identification background can be recognized quickly and accurately through the mobile terminal.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a method for quickly roll-calling and signing-in in a classroom based on deep learning face recognition technology. Background technique [0002] With the rapid development of the multimedia information age, accurate information acquisition, information representation and information processing have become indispensable research work and important technical means in today's scientific research. Face recognition technology has the advantages of good naturalness, non-invasiveness and high cost performance, which makes face recognition technology receive extensive attention and research in visual data processing. [0003] Traditional face recognition research is mainly based on the eigenface method, the face recognition algorithm Fisherface based on subspace analysis, the method based on elastic graph matching, SVD decomposition, face isodensity line analysis and matching, h...

AI Technical Summary

Problems solved by technology

However, face recognition technology has been applied to limited scenarios such as identification of public security systems, driver's licenses, monitoring systems of banks and customs, automatic door guard systems, unit attendance, and mobile phone unlocking. It has been widely promoted and has huge potential application value, which needs to be further explored

The roll call method of school teachers is still relatively traditional, which is not only time-consuming and laborious, but also greatly affects classroom order and efficiency

Images