Work

Implementasi Algoritma Facenet Dan Haversine Pada Absensi Karyawan Online Di Toko Merah Putih Medan

Year: 2026
Type: article
Abstract: Employee attendance at Toko Merah Putih Medan is recorded through photographs submitted to a WhatsApp group, which are then manually checked and recorded by an administrator. This study aims to implement FaceNet for facial verification and Haversine for location validation in an Android-based online employee attendance application. The application was developed using the Waterfall method, Flutter, Dart, and Firebase Realtime Database. Facial verification was performed by comparing embeddings using Euclidean distance, while location validation calculated the distance between the employee and the store’s reference point within a 10-meter radius. Testing showed that 10 same-person face pairs were accepted, with distances ranging from 0.340841 to 0.799974, while 10 different-person face pairs were rejected, with distances ranging from 1.014623 to 1.325277, using a threshold of 1.0. Location testing showed that distances of 6.7 and 9.9 meters were accepted, while a distance of 20.1 meters was rejected. All 35 black-box testing scenarios, comprising 17 administrator scenarios and 18 employee scenarios, produced the expected results. The application was also successfully installed and used on 10 Android devices. These results indicate that the application integrates facial verification, location validation, and attendance recording while supporting employee data management, store settings, recording of leave, sickness, and unexcused absences, attendance reports, and attendance history. The findings are limited to the samples and conditions tested.
Language: English
Open Access status: hybrid
OpenAlex ID: W7236414918
FWCI:
Cites: 19
Cited by:
Related to: 10