Scholarlink

In any educational institution, attendance plays a crucial role in deciding the overall performance of a student for that academic year. Generally, attendance in regular classes is taken manually either by calling out a student’s name or their roll number/ID. This is a traditional method followed by most institutions, which is time-consuming, prone to error, and proxy-marking. In this paper, we explore how AI addresses these limitations of current attendance systems. We have implemented an automated attendance-taking system using facial recognition technology. Here, we compared the performance of four facial-recognition models – Dlib, FaceNet, InsightFace (ArcFace), and DeepFace. InsightFace with ArcFace embeddings was identified as the most accurate under real classroom conditions. Real-time footage is processed by the system, and authentic records are exported to the database. Another important aspect of our paper is the implementation of behavioral intelligence modules to flag students in unauthorized lectures and anti-cheating surveillance. If any abnormal activity is detected in examination halls by analyzing the posture of the student, automated alerts are sent to the authority. In this way, we enhance security, transparency, and scalability, combining facial recognition with behavioral analytics for management in academic institutions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Implementation of Real-Time Computer Vision-Based System for Monitoring Attendance and Examination Misconduct

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Prarambhika Bhattacharjee,
  • Debolina Saha,
  • Shubhrakamal Saha,
  • Suhana Dogra,
  • Koushiki Ghosh,
  • Sudip Dogra

摘要

In any educational institution, attendance plays a crucial role in deciding the overall performance of a student for that academic year. Generally, attendance in regular classes is taken manually either by calling out a student’s name or their roll number/ID. This is a traditional method followed by most institutions, which is time-consuming, prone to error, and proxy-marking. In this paper, we explore how AI addresses these limitations of current attendance systems. We have implemented an automated attendance-taking system using facial recognition technology. Here, we compared the performance of four facial-recognition models – Dlib, FaceNet, InsightFace (ArcFace), and DeepFace. InsightFace with ArcFace embeddings was identified as the most accurate under real classroom conditions. Real-time footage is processed by the system, and authentic records are exported to the database. Another important aspect of our paper is the implementation of behavioral intelligence modules to flag students in unauthorized lectures and anti-cheating surveillance. If any abnormal activity is detected in examination halls by analyzing the posture of the student, automated alerts are sent to the authority. In this way, we enhance security, transparency, and scalability, combining facial recognition with behavioral analytics for management in academic institutions.