PERBANDINGAN OPTIMASI SGD DAN ADAM PADA ARSITEKTUR YOLOv5 (YOU ONLY LOOK ONCE) UNTUK DETEKSI ALAT PELINDUNG DIRI

Effendi, Dwi Wahyu (2022) PERBANDINGAN OPTIMASI SGD DAN ADAM PADA ARSITEKTUR YOLOv5 (YOU ONLY LOOK ONCE) UNTUK DETEKSI ALAT PELINDUNG DIRI. Undergraduate thesis, UPN Veteran Jawa Timur.

[img]
Preview
Text (cover)
18081010017.-cover.pdf

Download (5MB) | Preview
[img]
Preview
Text (bab1)
18081010017.-bab1.pdf

Download (106kB) | Preview
[img]
Preview
Text (bab2)
18081010017.-bab2.pdf

Download (477kB) | Preview
[img] Text (bab3)
18081010017.,-bab3.pdf
Restricted to Registered users only until 23 November 2024.

Download (144kB)
[img] Text (bab4)
18081010017.-bab4.pdf
Restricted to Registered users only until 23 November 2024.

Download (1MB)
[img]
Preview
Text (bab5)
18081010017.-bab5.pdf

Download (69kB) | Preview
[img]
Preview
Text (daftarpustaka)
18081010017.-daftarpustaka.pdf

Download (106kB) | Preview

Abstract

Personal Protective Equipment is a tool that must be used by all construction workers in K3. The use of PPE is intended to prevent health hazards or safety problems in the workplace. In this study, researchers conducted object detection for 4 classes, namely head, helmet, novest and vest. The researcher uses SGD and ADAM optimization on the YOLOv5 (You Only Look Once) architecture to perform object detection recognition. This research, started by conducting a literature study stage followed by data acquisition. The next stage is designing and training models using several types of YOLO including YOLOv5x, YOLOv5m, YOLOv5n. Each type of YOLO will be trained using two optimization algorithms, namely SGD and ADAM. After that the model will be tested with a data test, the test results will be evaluated and compared the level of accuracy of each model to find out the best YOLO model performance. The YOLOv5x optimizer SGD model gives the best performance with mAP@0.5 values ​​of 0.957 and mAP@0.5:.95 of 0.641, while YOLOv5x optimizer ADAM gets the lowest performance with values ​​of mAP@0.5 of 0.719 and mAP@0.5:.95 of 0.375.

Item Type: Thesis (Undergraduate)
Contributors:
ContributionContributorsNIDN/NIDKEmail
Thesis advisorAnggraeny, Fetty Tri0711028201fettyanggraeny.if@upnjatim.ac.id
Thesis advisorRizki, Agung Mustika0025079302agung.mustika.if@upnjatim.ac.id
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions: Faculty of Computer Science > Departemen of Informatics
Depositing User: Unnamed user with email 18081010017@student.upnjatim.ac.id
Date Deposited: 23 Nov 2022 06:45
Last Modified: 23 Nov 2022 06:45
URI: http://repository.upnjatim.ac.id/id/eprint/10390

Actions (login required)

View Item View Item