<![CDATA[Digitalisasi presensi staf pada fasilitas kesehatan primer diperlukan untuk meningkatkan akurasi, efisiensi administrasi, dan akuntabilitas kehadiran. Penelitian ini bertujuan merancang dan mengimplementasikan prototipe sistem presensi staf Puskesmas Kota Tais berbasis pengenalan wajah dengan mengintegrasikan algoritma Haar Cascade untuk deteksi wajah dan Local Binary Pattern Histogram (LBPH) untuk pengenalan identitas. Metode pengembangan yang digunakan adalah prototype dengan tahapan communication, quick plan, modeling, construction of prototype, serta deployment and feedback. Sistem dibangun berbasis web dengan dua peran pengguna, yaitu admin dan staf, serta didukung pipeline pengenalan wajah yang mencakup deteksi wajah, validasi kualitas citra dan deteksi mata sebagai lapisan anti-spoofing, pra-pemrosesan Tan-Triggs, Difference of Gaussian, dan CLAHE, serta pengenalan LBPH dengan konfigurasi radius 2, neighbors 12, grid_x 8, dan grid_y 8. Hasil pengujian menunjukkan bahwa sistem mampu mengenali wajah staf terdaftar pada kondisi pencahayaan normal dengan skor kepercayaan 85,27%–85,60%, melampaui threshold 75%. Pada skenario impostor, skor turun menjadi 39,43%, sehingga akses berhasil ditolak. Pengujian blackbox terhadap 22 skenario mencatat keberhasilan fungsional 100%. Temuan ini menunjukkan bahwa integrasi Haar Cascade dan LBPH layak digunakan sebagai solusi presensi digital yang responsif, contactless, dan akuntabel pada lingkungan Puskesmas. The digitization of staff attendance in primary healthcare facilities is necessary to improve recording accuracy, administrative efficiency, and attendance accountability. This study aimed to design and implement a face-recognition-based attendance prototype for staff at Tais City Public Health Center by integrating Haar Cascade for face detection and Local Binary Pattern Histogram (LBPH) for identity recognition. The system was developed using the prototype method, consisting of communication, quick planning, modeling, prototype construction, and deployment and feedback. The web-based system supports two user roles, namely administrator and staff, and applies a layered face-recognition pipeline that includes face detection, image-quality validation, eye detection as an anti-spoofing layer, Tan-Triggs preprocessing, Difference of Gaussian, CLAHE, and LBPH recognition with radius 2, neighbors 12, grid_x 8, and grid_y 8. The test results showed that the system recognized registered staff faces under normal lighting conditions with confidence scores of 85.27%–85.60%, exceeding the 75% acceptance threshold. In the impostor scenario, the confidence score dropped to 39.43%, and access was successfully rejected. Blackbox testing on 22 functional scenarios produced a 100% success rate. These findings indicate that the integration of Haar Cascade and LBPH is feasible as a responsive, contactless, and accountable digital attendance solution for a public health center environment. ]]>
Copyrights © 2026
