PENGEMBANGAN SISTEM PRESENSI CERDAS BERBASIS LOKASI DAN PENGENALAN WAJAH DENGAN DETEKSI PEMALSUAN LOKASI (FAKE GPS DETECTION) DI KANTOR DESA BONO

Abstract

Bono Village Office, Pakel District, Tulungagung Regency still uses a manual attendance system based on paper registers that is prone to data manipulation, proxy attendance, and recording inaccuracies, resulting in low levels of accountability and transparency in employee attendance. This study aims to develop a smart attendance system (E-Presensi) based on location and facial recognition with a Fake GPS Detection feature. The system was developed using the Prototype method with a three-layer architecture: React Native (Expo SDK 54) as the Android mobile interface, Laravel 10 as the backend REST API, and Python Flask as the AI microservice for Face Recognition (FaceNet, cosine threshold 0.40) and Liveness Detection (MediaPipe blink detection). Four security layers are executed sequentially: Fake GPS Detection, Geofencing (Haversine formula, 20-meter radius), Face Recognition, and Liveness Detection. Black Box Testing on 55 scenarios across 13 modules showed 100% successful results. Evaluation using a Likert-scale questionnaire administered to 5 employees of Bono Village Office obtained an average score of 4.60/5.00 (92.0%), classified as Very Good. System Usability Scale (SUS) evaluation obtained an average score of 86.0 (Grade B/Good). This system integrates four security layers simultaneously in a single smart attendance system for village governance, which has not been found in previous research. ABSTRAK Kantor Desa Bono, Kecamatan Pakel, Kabupaten Tulungagung masih menggunakan sistem presensi manual berbasis buku presensi yang rentan terhadap manipulasi data, titip absen, dan ketidakakuratan pencatatan, sehingga menyebabkan rendahnya akuntabilitas dan transparansi kehadiran pegawai. Penelitian ini bertujuan mengembangkan sistem presensi cerdas (E-Presensi) berbasis lokasi dan pengenalan wajah yang dilengkapi fitur Fake GPS Detection. Sistem dikembangkan menggunakan metode Prototype dengan arsitektur tiga lapis: React Native (Expo SDK 54) sebagai antarmuka mobile Android, Laravel 10 sebagai backend REST API, dan Python Flask sebagai AI microservice untuk Face Recognition (FaceNet, threshold cosine 0,40) dan Liveness Detection (MediaPipe blink detection). Empat lapisan keamanan dieksekusi secara berurutan: Fake GPS Detection, Geofencing (formula Haversine, radius 20 meter), Face Recognition, dan Liveness Detection. Black Box Testing terhadap 55 skenario pada 13 modul menunjukkan 100% hasil berhasil. Evaluasi kuesioner skala Likert terhadap 5 responden pegawai Kantor Desa Bono memperoleh nilai rata-rata 4,60/5,00 (92,0%) dengan kategori Sangat Baik. Evaluasi System Usability Scale (SUS) memperoleh nilai rata-rata 86,0 (Grade B/Good). Sistem ini mengintegrasikan empat lapisan keamanan sekaligus dalam satu sistem presensi cerdas untuk pemerintahan desa, yang belum ditemukan pada penelitian terdahulu.