Sistem Presensi Otomatis Menggunakan Pengenalan Wajah Berbasis Deep Learning dan Real-Time Database

The attendance system is a crucial component in the operations of any organization. However, most existing attendance systems still require significant time or manual intervention from users. This study aims to develop a deep learning-based face recognition application with a real-time database to record attendance automatically. This approach is expected to make the attendance process more accurate, faster, and more convenient compared to traditional attendance methods. The study employs a quantitative method through primary data analysis from laboratory testing using dummy data. This testing aims to measure the accuracy of the face recognition system in automatically recording attendance. A face recognition application prototype has been successfully developed with real-time database integration using the Python programming language. The test results show that the application can recognize all faces in the database with a very high accuracy level. The system performance metrics indicate an accuracy of 99.1%, precision of 98.7%, recall of 98.7%, and F1-score of 98.7%. Additionally, the model has been implemented on an NVIDIA Jetson Nano mini-processor, demonstrating efficient operation on low-power hardware and real-time face recognition with optimal processing speed.