Face Recognition Based Attendance System in the Raspberry Pi 4

The development and deployment of the proposed con- tactless attendance system demonstrate the practical viability of using deep learning-based facial recognition on low-cost embedded platforms like the Raspberry Pi 4. By integrating a CNN-based face recognition model with OpenCV, the sys- tem achieved a high recognition accuracy of over 96%, sig- nificantly outperforming traditional HOG-based approaches, which reached approximately 90%. The system maintained reliable performance under varying lighting conditions and facial orientations, ensuring robust real-time detection and recognition. Furthermore, the solution successfully automated attendance tracking by logging data into structured Excel sheets while preventing duplicate entries for the same user within a day. Its fully offline and scalable nature makes it highly suitable for academic institutions and organizational environments, especially where cloud connectivity may be limited. The optimized inference time and resource-efficient design further confirm the system’s effectiveness for real- time use. Overall, the implementation fulfills the key goals of automation, accuracy, hygiene, and cost-effectiveness, pro- viding a strong foundation for future enhancements such as cloud integration and multi-user management. Although the system performs reliably for small to medium scale deployments, several enhancements could be considered to extend its scalability and functionality. Integrating a real-time graphical user interface (GUI) could improve administrator interaction and visibility. Additionally, incorporating cloud- based synchronization would enable centralized data access and long-term archival. Future work may also explore support for masked face detection, thermal camera integration for temperature screening, and mobile-based access for remote attendance verification [23] [24].

[1]     S. Gupta and P. Mehta, “Deepfacesecure: Privacy-enhanced facial recog- nition framework for smart campus,” in Proceedings of the 2024 IEEE International Conference on Smart Computing (SMARTCOMP), 2024, pp. 89–96. [2]     T. Bose and R. Sinha, “Optimized raspberry pi-based smart attendance system using cnn,” in Proceedings of the 2023 International Conference on Embedded AI Systems (ICEAIS), 2023, pp. 101–106. [3]     M. Elbaz, “Cnn-hog hybrid models for enhanced facial emotion and identity detection,” Journal of Real-Time Image Processing, vol. 20, pp. 456–470, 2023. [4]     D. E. King, “Dlib-ml: A machine learning toolkit,” Journal of Machine Learning Research, vol. 10, pp. 1755–1758, 2009, accessed: June 25, 2025. [Online]. Available: http://dlib.net [5]     OpenCV Team, “Opencv documentation,” https://docs.opencv.org/, 2024, accessed: June 30, 2025. [6]     S. Sivakumar, T. J. John, G. T. Selvi, B. Madhu, C. U. Shankar, and K. Arjun, “Iot based intelligent attendance monitoring with face recog- nition scheme,” in 2021 5th International Conference on Computing Methodologies and Communication (ICCMC). IEEE, 2021, pp. 349– 353. [7]     Y. Jang, “Real-time biometric surveillance using low-power edge ai,” IEEE Transactions on Industrial Informatics, 2025, early Access. [Online]. Available: https://ieeexplore.ieee.org/document/XXXXX [8]     Y. Wang et al., “A survey of face recognition systems in edge ai: Recent advances and challenges,” IEEE Transactions on Artificial Intelligence, vol. 4, no. 1, pp. 12–29, 2023. [9]     Raspberry Pi Foundation, “Raspberry pi 4 model b datasheet,” https://www.raspberrypi.com/products/raspberry-pi-4-model-b/, 2024, accessed: June 18, 2025. [10]    D. Fernandez et al., “Comparative analysis of face recognition models for offline attendance monitoring,” Sensors, vol. 23, no. 12, pp. 5234– 5249, 2023.