Amri, Saeful (2023) IMPELEMENTASI DEEP LEARNING YOU ONLY LOOK ONCE v5 PADA PENGENALAN WAJAH SECARA REAL-TIME. Skripsi thesis, Universitas Muhammadiyah Sukabumi.

	Text LEMBAR PENGESAHAN PEMBIMBING.pdf Download (788kB)
	Text HALAMAN PERNYATAAN PLAGIARISME.pdf Download (680kB)
	Text BAB I.pdf Download (1MB)
	Text BAB II.pdf Restricted to Repository staff only Download (2MB) | Request a copy
	Text BAB III.pdf Restricted to Repository staff only Download (990kB) | Request a copy
	Text BAB IV.pdf Restricted to Repository staff only Download (2MB) | Request a copy
	Text BAB V.pdf Download (765kB)

Abstract

Every place has a security system. In an effort to improve security, be it the government, universities, shops, offices, and others, they must have installed CCTV (Closed Circuit Television) as a surveillance tool. At the Muhammadiyah University of Sukabumi there is a security problem where there is concern about the level of security that has not been maximized on campus because of the large number of students and the increase in the number of students every year making it difficult for security officers to verify whether people in the campus area are UMMI students or not, especially when they wearing a helmet making it difficult to identify them. To overcome this and help improve performance in surveillance, a biometric authentication system is used to validate a person's face by monitoring the traffic of students or foreigners with surveillance cameras equipped with Artificial Intelligence. The purpose of this research is to increase the level of security in the scope of the UMMI campus through the implementation of Deep Learning YOLOv5 on real-time facial recognition by building a security system that is able to recognize someone through him and verify their identity by utilizing artificial intelligence. The research method involves using the YOLOv5 Deep Learning algorithm, FaceNet, and Support Vector Machine (SVM) to train a facial recognition system by distinguishing between students and non-students based on their facial features. Yolov5 aquariums used the WIDEFACE filtered data set for small faces, achieving a mean precision (mAP) of 0.838 at an IoU park of 0.5 and a mAP of 0.498 at an IoU park of 0.5:0.95. SVM training resulted in a precision of 0.90, recall of 0.90, and a 90% F1 score based on the average weight. By testing using 99 images of data for each student, the system can already detect and recognize someone with 89% results for a distance of 1 meter and 53% for a distance of 2 meters. The results showed that the detection of the YOLOv5 Deep Learning algorithm on the Face Recognition system can effectively increase the level of security on campus by recognizing students' faces in real-time quite accurately. Utilizing this system will provide a safer and more peaceful environment for students, staff, and visitors, so that UMMI's good name is maintained.

Item Type:	Thesis (Skripsi)
Uncontrolled Keywords:	Video Surveillance, Face Recognition, YOLOv5, FaceNet, Support Vector Machine
Subjects:	T Technology > T Technology (General)
Divisions:	Fakultas Sains dan Teknologi > Teknik Informatika
Depositing User:	Perpus ID UMMI
Date Deposited:	02 Aug 2023 06:39
Last Modified:	02 Aug 2023 06:39
URI:	http://eprints.ummi.ac.id/id/eprint/3268

Actions (login required)

View Item