Pengembangan Sistem Presensi Otomatis Menggunakan Yolo dan Facenet

sani, Muhammad Fadhly (2025) Pengembangan Sistem Presensi Otomatis Menggunakan Yolo dan Facenet. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6022231075-MASTER_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (9MB) | Request a copy

Abstract

Implementasi sistem presensi otomatis berbasis pengenalan wajah di lingkungan dinamis menghadapi tantangan seperti bentuk ruangan, konfigurasi tempat duduk, posisi peralatan, dan aktivitas di ruangan yang dapat menghalangi pandangan kamera terhadap wajah peserta. Kondisi tersebut menyebabkan penggunaan satu kamera menjadi tidak efektif karena keterbatasan cakupan pandang. Untuk mengatasi permasalahan ini, penelitian ini mengusulkan pendekatan multi-kamera dengan penentuan posisi dan orientasi kamera secara optimal menggunakan algoritma Particle Swarm Optimization (PSO). PSO bekerja berdasarkan hasil klasifikasi identitas dari berbagai konfigurasi kamera, yang diperoleh melalui tahapan pemrosesan citra berurutan, yakni peningkatan kontras rekaman menggunakan CLAHE, deteksi wajah menggunakan YOLOv11, peningkatan kualitas wajah dengan GFPGAN, dan ekstraksi fitur wajah menggunakan FaceNet512. Identitas wajah ditentukan dengan metode cosine similarity dengan threshold ≥ 0,6. Hasil pengujian menunjukkan bahwa sistem berhasil mengenali seluruh peserta dalam ruangan secara otomatis dengan posisi dan orientasi kamera yang telah ditentutkan oleh PSO dengan jumlah kamera tersedikit. Mampu mencatat kehadiran peserta dengan tingkat keberhasilan 100%.
===================================================================================================================================
The implementation of face recognition-based automatic attendance systems in dynamic environments faces challenges such as room shape, seating configuration, equipment position, and activities in the room that can obstruct the camera’s view of the participant’s face. These conditions cause the use of a single camera to be ineffective due to limited visibility. To overcome this problem, this research proposes a multi-camera approach with optimal camera positioning and orientation using the Particle Swarm Optimization (PSO) algorithm. PSO works based on the identity classification results of various camera configurations, which are obtained through sequential image processing stages, namely, recording contrast enhancement using CLAHE, face detection using YOLOv11, face quality enhancement with GFPGAN, and face feature extraction using FaceNet512. Facial identity is determined by the cosine similarity method with a threshold ≥ 0.6. The test results show that the system successfully recognizes all participants in the room automatically with the position and orientation of the camera that has been determined by PSO with the least number of cameras. Able to record attendance of participants with 100% success rate.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	presensi otomatis, pengenalan wajah, yolov11, gfpgan, facenet, clahe cosine similarity theorem, particle swarm optimization, automatic attendance, face recognition, yolov11, facenet, clahe, cosine similarity theorem, particle swarm optimization
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.62 Decision support systems T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7882.P3 Pattern recognition systems
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User:	Muhammad Fadhly Sani
Date Deposited:	22 Jul 2025 05:43
Last Modified:	22 Jul 2025 05:43
URI:	http://repository.its.ac.id/id/eprint/120032

Actions (login required)

View Item