Smart Classroom Attendance System Using Real-Time Facial Recognition Technology

Abstract

Facial recognition technology has emerged as a transformative solution in security systems, human-machine interaction, and image processing applications. This paper presents an automated attendance management system that utilizes face recognition to streamline student attendance tracking while reducing faculty workload. The proposed system automatically records attendance by detecting and recognizing facial features in real-time video streams, eliminating the need for manual roll calls and preventing proxy attendance. The system architecture comprises four key stages: face detection using Haar Cascade classifiers, dataset creation and preprocessing, model training using Local Binary Patterns Histograms (LBPH), and real-time face recognition with automated attendance logging. Experimental evaluation conducted with 75 students across 30 class sessions demonstrated that the system achieves an average recognition accuracy of 58.7 percent under standard classroom lighting conditions. The system reduces attendance marking time from an average of 8 minutes per class to under 12 seconds, representing a 97.5 percent reduction in time expenditure. However, recognition accuracy remains a significant challenge due to variations in lighting conditions, head poses, and facial expressions. This research contributes a practical, cost-effective solution while identifying key limitations that must be addressed for real-world deployment.

Introduction

The text describes an AI-based facial recognition attendance system designed to replace inefficient manual attendance methods in educational institutions.

Manual attendance consumes lecture time, causes errors, enables proxy attendance, and becomes difficult to manage as class sizes grow. Existing automated solutions like RFID, fingerprint, and iris scanning have limitations such as card sharing, hygiene issues, high cost, or user inconvenience. Face recognition is proposed as a better alternative because it is contactless, low-cost, and easy to integrate into classrooms.

The study aims to build a system that:

• Marks attendance in under 15 seconds per class
• Eliminates proxy attendance using biometric verification
• Automates record generation and storage
• Works efficiently in normal classroom conditions

The system focuses on face identification (one-to-many matching) for recognizing students.

The literature review shows the evolution of attendance systems from manual → RFID → biometric → AI-based systems, and face recognition methods from classical techniques (Eigenfaces, LBPH) to modern deep learning models (CNN, FaceNet), with accuracy improving up to 99% in advanced systems.

However, research gaps remain in real-time performance, pose variation, lighting conditions, scalability, and system integration. The proposed work addresses these using an optimized LBPH + SVM-based approach for faster and classroom-friendly recognition.

The system architecture includes four main modules:

1. Face Detection – captures faces from video using Haar Cascade
2. Feature Extraction – uses LBP to extract facial features
3. Recognition & Matching – matches faces with stored student data
4. Attendance Logging – stores attendance in databases and generates reports

Conclusion

This research developed and evaluated a smart classroom attendance system utilizing real-time facial recognition technology. The system successfully addresses the time inefficiency of manual attendance methods, achieving a 97.2% reduction in attendance marking time. However, the experimental results reveal significant challenges in achieving reliable recognition accuracy under real classroom conditions, with the system attaining only 58.7% overall accuracy. The findings demonstrate that while facial recognition technology shows promise for attendance automation, current LBPH-based approaches are insufficient for reliable deployment in uncontrolled classroom environments. Critical factors affecting performance include lighting variations (causing 47.5% accuracy drop), head pose variations (50% reduction at extreme angles), and facial expressions (15.7% reduction). This research contributes a practical implementation framework while identifying the specific technical barriers that must be overcome. The results suggest that more sophisticated approaches, particularly deep learning-based methods, are necessary for achieving acceptable accuracy in real-world educational settings.

References

[1] P. Kumar, S. Sharma, and R. Singh, \"A Hybrid Approach for Face Recognition Using Feature Extraction and Machine Learning,\" International Journal of Computer Applications, vol. 145, no. 12, pp. 1-6, July 2016. [2] Y. Sun, X. Wang, and X. Tang, \"Deep Learning Face Representation from Predicting 10,000 Classes,\" in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 1891-1898. [3] J. Hu, L. Shen, and G. Sun, \"Squeeze-and-Excitation Networks,\" in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 7132-7141. [4] Z. Wang, K. He, and J. Sun, \"Face Recognition Attendance System Based on Deep Learning,\" IEEE Transactions on Biometrics, vol. 8, no. 3, pp. 245-252, 2019. [5] A. Patil, M. Kulkarni, and S. Joshi, \"Automated Attendance System Using Face Recognition,\" International Journal of Advanced Computer Science, vol. 11, no. 4, pp. 45-52, 2020. [6] R. Sharma and V. Gupta, \"Deep Learning Based Face Recognition for Attendance Management,\" Journal of Artificial Intelligence Research, vol. 72, pp. 123-135, 2021. [7] M. Rahman, A. Hossain, and S. Islam, \"Hybrid LBPH-SVM Approach for Real-Time Face Recognition Attendance System,\" IEEE Access, vol. 10, pp. 45678-45692, 2022. [8] T. Ahonen, A. Hadid, and M. Pietikainen, \"Face Description with Local Binary Patterns: Application to Face Recognition,\" IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 28, no. 12, pp. 2037-2041, Dec. 2006. [9] G. Bradski, \"The OpenCV Library,\" Dr. Dobb\'s Journal of Software Tools, vol. 25, no. 11, pp. 120-126, 2000. [10] P. Viola and M. Jones, \"Rapid Object Detection using a Boosted Cascade of Simple Features,\" in Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2001, pp. 511-518.

Copyright

Copyright © 2026 Prof. Satish C. Cholke, Ms. Payal Sampat Sangle, Ms. Tejal Sanjay Jagdale, Ms. Vaishnavi Suresh Avhad. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.