Skripsi

PENDETEKSIAN SAMPAH BAWAH AIR MENGGUNAKAN METODE YOU ONLY LOOK ONCE DENGAN INTEGRASI EFFICIENT MULTI-SCALE ATTENTION

---

Plastic waste in aquatic ecosystems is a pressing global issue; however, underwater monitoring is often constrained by environmental conditions such as limited lighting and the small visual appearance of objects. Although the You Only Look Once (YOLO) algorithm excels in real-time detection, it still faces limitations in accurately detecting small objects. Therefore, this study aims to implement YOLOv8 integrated with the Efficient Multi-Scale Attention (EMA) module to improve the accuracy of underwater waste detection, particularly for small-sized objects. In the architectural development, the EMA module is integrated into the C2f block within the backbone to form a modified C2f_EMA block, which is designed to enhance multi-scale feature extraction and focus the network’s attention on salient features without significantly increasing computational overhead. The dataset used in this study is a combination of the Trash Dataset ICRA 2019 and extracted underwater video frames. Experimental results show that the YOLOv8+EMA model with a batch size of 16 and a learning rate of 0.01 achieved a Precision of 0.765, a Recall of 0.584, an mAP@50 of 0.654, and an mAP@50–95 of 0.398, demonstrating a consistent performance improvement over the baseline YOLOv8 model while maintaining model efficiency.

---

Availability

Detail Information

Series Title

-

Call Number

T1930022026

Publisher

Indralaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2026

Collation

xx, 143 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

006.407

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Teknik Informatika
Deteksi Sampah Bawah Air

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment