Pengembangan Sistem Absensi Karyawan Berbasis Pengenalan Wajah Menggunakan Convolutional Neural Network (CNN)

Abstract

Employee attendance systems require an effective mechanism for recording attendance to support employee attendance management in companies. Face recognition using a Convolutional Neural Network (CNN) is one of the technologies that can be applied to identify faces from digital images. This study aims to develop an employee attendance system based on face recognition using the CNN algorithm. The research data consist of employee facial images collected manually. System development includes CNN model training and the implementation of several features, including clock-in, clock-out, leave requests, permission requests, and attendance history accessible to administrators. Model training was conducted for 50 epochs using Google Colab, while the application was evaluated through manual testing. The results indicate that the model achieved accuracy values of 98.50% and 99.00% at epoch 50, as reported in the training results. The developed system also provides features for managing employee attendance records and viewing attendance history. Therefore, the system has the potential to support the digitalization of employee attendance recording in companies. However, the specific accuracy metrics and evaluation data split need to be described in greater detail to support a more precise interpretation of the model's performance.