Aero Face: Biometric Attendance & Administration System

Abstract

Even though most colleges and exam centers in Andhra Pradesh still use traditional attendance methods, such as paper registers or card machines, these approaches run into real problems in practice. Staff attendance is often marked as present by a colleague, even when that person is not actually there. Because most lab environments are dusty, fingerprint scanners jam frequently. And cloud-based systems go down whenever the institution\'s network goes down. In this paper, we propose a novel attendance system called AeroFace, which uses the webcam already available on the lab PC to capture images of employees, runs face recognition in the browser, and stores all data within the institution\'s local network, with no cloud dependency, no dedicated hardware, and no internet connection requirement after initial setup. The recognition engine uses face-api.js, which runs FaceNet-based neural network models through TensorFlow.js on the browser\'s WebGL-accelerated GPU pipeline. During enrollment, a 128-dimensional descriptor vector is extracted from an employee\'s photograph and stored; during live scanning, the same pipeline processes the employee\'s webcam feed, and a Euclidean-distance comparison is made against the stored descriptor vectors. The result of the comparison is logged along with information such as GPS location, a photo snapshot of the employee, and a confidence score. The back-end of the system is a 34-endpoint Node.js and Express API, which supports 12 administrative modules, including attendance logs, employee management, exam-slot booking for up to 120 systems, an income and expense ledger, salary computation with printable payslips, travel planning, and supervisor approval of transactions. The system was validated in a college laboratory environment through 25 functional test cases, divided into unit, integration, and multi-client network tests. The total size of the recognition pipeline is approximately 6.5 MB, and it requires no internet connection at runtime. This shows that a browser-native, deep-learning-based recognition system is a practical, low-cost, and hardware-independent alternative to commercial biometric attendance systems.

Introduction

AeroFace is a browser-based facial recognition attendance system designed for colleges and examination centers to replace manual attendance registers and expensive fingerprint biometric systems. Traditional attendance methods are slow, error-prone, and susceptible to proxy attendance, while fingerprint systems require costly hardware, frequent maintenance, and reliable internet connectivity. AeroFace overcomes these limitations by using existing webcams and local area networks (LANs), performing facial recognition directly in the browser through face-api.js and TensorFlow.js, with no dependence on cloud services or specialized biometric devices.

The system is built on a two-tier architecture consisting of a browser-based recognition layer and a Node.js/Express server with 34 RESTful APIs backed by a JSON database. During enrollment, facial features are converted into 128-dimensional descriptors and stored locally. During attendance, live webcam images are matched with stored descriptors using Euclidean distance, while recording GPS coordinates, timestamps, confidence scores, IP address, and photo evidence. AeroFace also includes role-based access control, exam slot booking, financial management, payroll generation, travel planning, and automated database backup, all functioning completely offline on a LAN.

The system was tested at AQJ College for PG Studies, Visakhapatnam, where 25 functional test cases covering authentication, face recognition, attendance logging, financial modules, role-based access, backup, and multi-client operation were successfully completed. The results demonstrate that AeroFace provides a reliable, secure, low-cost, and tamper-resistant attendance solution that eliminates dependence on internet connectivity while integrating attendance, administration, payroll, and examination management into a single platform.

Conclusion

AeroFace demonstrates that browser-native deep learning inference, built on face-api.js and TensorFlow.js, can deliver a hardware-independent biometric attendance solution suitable for colleges and examination centers. By coupling FaceNet-based recognition with a self-hosted Node.js/Express REST API and a twelve-module administrative dashboard, the system consolidates attendance, finance, payroll, booking, and logistics functions that are traditionally fragmented across manual registers — while operating entirely offline on a local network. The result is a cost-effective, low-maintenance alternative to commercial biometric hardware, one that remains fully functional without continuous internet connectivity. A notable outcome of this work is how much administrative overhead could be consolidated into a single platform: attendance registers, expense sheets, salary calculations, booking logs, and approval chains that were previously maintained separately are now handled through one dashboard, backed by a single JSON data store that can be backed up and restored in one step.

References

[1] F. Schroff, D. Kalenichenko, and J. Philbin, \"FaceNet: A Unified Embedding for Face Recognition and Clustering,\" in Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), 2015. [2] V. Muhler, \"face-api.js: JavaScript Face Recognition API for the Browser and Node.js Implemented on top of tensorflow.js core,\" GitHub Repository, 2018. [3] M. Turk and A. Pentland, \"Eigenfaces for Recognition,\" Journal of Cognitive Neuroscience, vol. 3, no. 1, pp. 71–86, 1991. [4] R. T. Fielding, \"Architectural Styles and the Design of Network-based Software Architectures,\" Ph.D. dissertation, Univ. of California, Irvine, 2000. [5] N. Sinai, \"TensorFlow.js: Machine Learning for the Web and Beyond,\" in Proc. Systems for Machine Learning Workshop, NeurIPS, 2018. [6] Node.js Foundation, \"Node.js Documentation v20 LTS,\" 2023. [Online]. Available: https://nodejs.org/en/docs [7] Express.js Team, \"Express.js 4.x API Reference,\" 2023. [Online]. Available: https://expressjs.com/en/4x/api.html [8] P. N. Belhumeur, J. P. Hespanha, and D. J. Kriegman, \"Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection,\" IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 19, no. 7, pp. 711–720, 1997. [9] Y. Taigman, M. Yang, M. Ranzato, and L. Wolf, \"DeepFace: Closing the Gap to Human-Level Performance in Face Verification,\" in Proc. IEEE CVPR, 2014, pp. 1701–1708. [10] MDN Web Docs, \"MediaDevices.getUserMedia(),\" Mozilla Developer Network, 2024. [11] W3C Working Group, \"WebGL 2.0 Specification,\" World Wide Web Consortium, 2021. [12] Ministry of Electronics and Information Technology, Government of India, \"Digital Personal Data Protection Act, 2023,\" Gazette of India, 2023.

Copyright

Copyright © 2026 Mycharla Madhavkumar, Tulasi Miryala. 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.