Face Recognition Attendance Management System

Abstract

Traditional attendance management in academic institutionsreliesonmanualroll-callsandpaperrecords,leading to significant time wastage, proxy attendance fraud, and data entryerrors.Thispaperpresentsafullyautomated,real- time Face Recognition Attendance Management System that eliminates manual intervention through biometric verification. The system implements a secure admin login module, section- wise student registration using unique roll numbers, and an adaptive 7-second timeout mechanism for efficient classroom capture. Built on OpenCV and dlib libraries, the system utilizes HistogramofOrientedGradients(HOG)forfacedetection and 128-dimensional deep embeddings for recognition, matching live faces against enrolled students using Euclidean distance. A Tkinter-based graphical interface enables faculty to operate the system without technical expertise, while pandas and openpyxl handleautomaticgenerationofdate-wiseExcelattendancesheets. The adaptive timeout algorithm refreshes the capture window when new faces are detected and automatically closes thecamera after 7 seconds of inactivity, optimizing classroom time. Experimentalevaluationdemonstratesrobustperformanceunder varied lighting conditions, successful duplicate prevention, and 100% accurate Excel export functionality. The system effectively prevents proxy attendance through biometric verification and provides structured digital records for administrative review.

Introduction

The Face Recognition Attendance Management System is designed to automate attendance tracking in educational institutions, addressing the limitations of traditional methods such as manual roll calls, paper-based records, and proxy attendance. Conventional systems consume valuable lecture time, are prone to human errors, and lack efficient digital record management. Existing RFID and fingerprint-based systems require specialized hardware and physical contact, while many computer vision solutions lack practical features such as section-wise organization, automated reporting, and user-friendly interfaces.

The proposed system uses Python, OpenCV, dlib, ResNet-34, and Tkinter to provide a secure, contactless, and automated attendance solution. It includes admin authentication, roll-number-based student registration, and an adaptive 7-second camera timeout that extends scanning when new faces are detected and automatically closes when no activity is observed, reducing classroom time consumption. Attendance records are automatically stored in date-wise Excel files, duplicate attendance entries are prevented, and unregistered individuals are rejected.

The system follows a modular workflow that begins with launching the application through an Anaconda environment, followed by secure admin login, section selection, student enrollment, attendance capture, and automated Excel report generation. During enrollment, student face images are converted into 128-dimensional facial embeddings and stored for future recognition. The recognition pipeline consists of HOG-based face detection, facial landmark alignment, feature embedding generation using ResNet-34, Euclidean distance matching, and validation to reject unknown faces and duplicate entries.

The implementation was evaluated using 25 registered students across multiple sections on a standard computer with a webcam. Experimental results achieved approximately 95.6% recognition accuracy, 100% duplicate prevention, 100% Excel export success, reliable unknown-face rejection, and average detection times of 80–120 milliseconds per frame. The system performed well under standard conditions, moderate lighting changes, and slight side poses, while successfully handling late arrivals and automatically managing camera activity through the adaptive timeout mechanism.

Key advantages of the system include secure administrator access, organized section-wise student management, automated attendance reporting, intelligent camera control, and a simple graphical interface that requires no programming knowledge. However, the current implementation has limitations, including the absence of liveness detection (making it vulnerable to photo spoofing), reduced accuracy under poor lighting or extreme side poses, support for only a single camera, and reliance on local storage rather than cloud-based databases.

Future enhancements include integrating anti-spoofing (liveness detection), migrating to cloud databases such as Firebase or PostgreSQL, supporting multi-camera environments for large classrooms, and integrating with Learning Management Systems (LMS) such as Moodle or Blackboard. Overall, the proposed system provides a practical, efficient, and deployment-ready attendance solution that significantly reduces manual effort while improving accuracy, security, and administrative efficiency.

Conclusion

This paper presented a practical, implemented Face Recog- nition Attendance Management System that addresses the critical deficiencies of traditional manual methods. By inte- gratingsecureadminauthentication,roll-number-basedsection management, and an innovative adaptive 7-second timeout mechanism,thesystemachievesefficientclassroomattendance capture while eliminating proxy attendance through biometric verification. TheautomatedExcelgenerationandTkinter-basedinterface make the system immediately deployable in educational envi- ronments without requiring specialized hardware or technical expertise.Experimentalresultsdemonstrate95.6%recognition accuracy,100%duplicateprevention,andsuccessfuloperation under varied classroom conditions. The system effectively re- duces attendance marking time from several minutes to under oneminutepersession,providingacost-effective,contact-free solution for smart classroom deployment.

References

[1] F. Deravi, “A survey of RFID based attendance monitoring systems,”International Journal of Computer Applications, vol. 121, no. 17, pp.12–17, 2015. [2] A.K.Jain,L.Hong,andS.Pankanti,“Biometricidentification,”CommunicationsoftheACM,vol.43,no.2,pp.90–98,Feb.2000. [3] P.ViolaandM.J.Jones,“Rapidobjectdetectionusingaboostedcascadeof simple features,” in Proceedings of the 2001 IEEE Computer SocietyConferenceonComputerVisionandPatternRecognition(CVPR2001),Kauai, HI, USA, Dec. 2001, vol. 1, pp. I-511–I-518. [4] A. S. Lateef et al., “Facial recognition technology-based attendancesystem,” Iraqi Journal for Computer Science and Mathematics, vol. 4,no. 3, 2023. [5] N. Dalal and B. Triggs, “Histograms of oriented gradients for humandetection,” in Proceedings of the IEEE Conference on Computer VisionandPatternRecognition(CVPR2005),SanDiego,CA,USA,2005,vol.1, pp. 886–893. [6] F.Schroff,D.Kalenichenko,andJ.Philbin,“FaceNet:Aunifiedembed-ding for face recognition and clustering,” in Proceedings of the IEEEConferenceonComputerVisionandPatternRecognition(CVPR2015),2015, pp. 815–823. [7] H. Dang, “Efficient deep learning approach for facial recognition usingan improved FaceNet model based on MobileNetV2,” 2023. [8] S. Vermaet al., “Automated attendance monitoring system using facialrecognition technology,” 2023. [9] C. Horn Boe, K. Ng, S. Haw, P. Naveen, and E.AbdulwahabAnaam,“An automated face detection and recognition for class attendance,”JOIV: International Journal on Informatics Visualization, vol. 8, no. 3,pp.1146–1153,Sep.2024. [10] J.Viswanathanetal.,“Smartattendancesystemusingfacerecognition,”EAI Endorsed Transactions on Scalable Information Systems, vol. 11,no. 5, Feb. 2024. [11] R. Shyam, A. Mishra, A. Kumar, A. Chowdhary, and A. K. Srivas-tava, “Recording of class attendance using DL-based face recognitionmethod,” in Data Science and Applications, Lecture Notes in Networksand Systems, vol. 818, Springer, Singapore, 2024.

Copyright

Copyright © 2026 Prof. Ashwini P. Ghatol, Amar G. Dahiwade, Yogendra N. Ninghot, Vicky V. Chavhan, Bhushan P. Shankarpure, Ajay M. Chavhan . 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.