Student Attendance Monitoring via Facial Recognition

Overview

Manual attendance marking is time-consuming and prone to proxy fraud. This project implements an automated student attendance monitoring system using deep learning-based facial recognition, enabling real-time, contactless attendance logging from classroom camera feeds.

Motivation

• Manual roll calls consume 5–10 minutes of every lecture
• Proxy attendance is a persistent problem in engineering colleges
• Post-pandemic, contactless solutions gained added relevance
• Results need to integrate with college ERP/MIS systems

System Design

• Face detection — MTCNN / Haar Cascade to detect and localize faces in classroom images
• Face recognition — deep CNN (FaceNet / VGGFace) to identify students from a registered database
• Attendance logging — automatic marking with timestamp, cross-referenced against class schedule
• Database integration — structured storage in MySQL/PostgreSQL
• Dashboard — faculty view for attendance reports and anomaly alerts

Key Contributions

• End-to-end pipeline from live camera feed to attendance database
• Handles partial occlusion, varying lighting, and multiple faces per frame
• Evaluated on real classroom data from an engineering college environment

Publication

Published in International Conference on Innovative Computing and Communications (ICICC 2022), Vol. 3, pp. 613–620, Springer Nature Singapore.

(Wankhade et al., 2022)

Status

Published — Springer ICICC 2022.