Skripsi

IMPLEMENTASI METODE YOU ONLY LOOK ONCE (YOLOV8) DALAM MENDETEKSI ACNE VULGARIS

---

This study aims to implement the You Only Look Once (YOLOv8) method for detecting acne in acne vulgaris images. Automatic detection of acne types is essential to support rapid, accurate, and efficient dermatological diagnosis. The research involves designing a YOLOv8-based detection model, which includes the stages of collecting and annotating facial image datasets, preprocessing, model training with various batch size and epoch configurations, and performance evaluation using precision, recall, and mean Average Precision (mAP) metrics. Based on the experimental results, the model with a batch size of 16 and 100 epochs achieved the best performance compared to other trained models, with a precision of 0.629, recall of 0.594, mAP@50 of 0.624, and mAP@50-95 of 0.289. These results indicate that dataset quality, particularly in the annotation and preprocessing stages, significantly influences the model’s detection performance. Therefore, the YOLOv8 method shows potential in assisting the automatic identification of acne lesions, although further improvement in data quality and augmentation strategies is still required to achieve more optimal results.

---

Availability

Detail Information

Series Title

-

Call Number

T1858712025

Publisher

Indralaya : Prodi Sistem Komputer, Fakultas Ilmu Komputer Universitas Sriwijaya., 2025

Collation

xiii, 90 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

616.530 7

Content Type

Text

Media Type

unmediated

Carrier Type

other (unmediated)

Edition

-

Subject(s)

Prodi Sistem Komputer
Deteksi Jerawat

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment