PENGENALAN INDIVIDU MELALUI IDENTIFIKASI WAJAH MENGGUNAKAN METODE YOU ONLY LOOK ONCE (YOLOv5)

(Individual Recognition Through Face Identification Based On You Only Look Once (YOLOv5) Method)

  • Nurul Hidayat Institut Teknologi Sepuluh Nopember, Kampus Keputih-Sukolilo, Surabaya
  • Suhud Wahyudi Institut Teknologi Sepuluh Nopember, Kampus Keputih-Sukolilo, Surabaya
  • Athaya Aufa Diaz Institut Teknologi Sepuluh Nopember, Kampus Keputih-Sukolilo, Surabaya

Abstract

Biometrics is a branch of applied mathematics to identify individuals who take advantage of the identification of the characteristics or special characteristics it has. Current system Individual recognition through facial identification is one of the biometric technologies which is popular. Despite its popularity, facial recognition systems have its own complexity in the identification process. As the times progressed, the system required to be able to recognize faces in real time. You Only Look Once (YOLO) is an object detection system that is famous for its ability to detect objects in real time. What’s more, the fifth version of YOLO or YOLOv5 does not require large memory. The best YOLOv5 model training results were obtained from taking parameters epoch=250, batch size=16, and learning rate=0.01. The testing process has been carried out on sixteen videos of which eight videos contain one different student in each video and the other eight videos contain two different students in each video. The testing process is also done by increasing and decreasing the brightness of the video. The best average accuracy result is 99.88% which is shown by testing on a video containing one student with a normal brightness level. 


Keywords: Biometrics, Face Identification, Individual Recognition, YOLOv5

Published
2022-08-14
How to Cite
HIDAYAT, Nurul; WAHYUDI, Suhud; DIAZ, Athaya Aufa. PENGENALAN INDIVIDU MELALUI IDENTIFIKASI WAJAH MENGGUNAKAN METODE YOU ONLY LOOK ONCE (YOLOv5). UNEJ e-Proceeding, [S.l.], p. 85 - 98, aug. 2022. Available at: <https://jurnal.unej.ac.id/index.php/prosiding/article/view/33497>. Date accessed: 03 dec. 2022.