Skripsi

IMPLEMENTASI OBJECT DETECTION BERBASIS YOU ONLY LOOK ONCE UNTUK IDENTIFIKASI FOREIGN OBJECT PADA SISTEM CONVEYOR BELT

---

This study aims to develop a deep learning-based object detection system using the YOLOv11n algorithm to identify foreign objects on coal conveyor belt systems. The study is motivated by the limitations of manual inspection methods in maintaining detection consistency and accuracy within mining environments characterized by high visual complexity, such as dust, uneven illumination, motion blur, and similarities in texture between materials. The dataset used in this research is the DsCGF Anhui–Guobei productive state subset. The research stages include data pre-processing consisting of label conversion, image enhancement, cleaning, and dataset balancing, followed by model training using various parameter configurations. The results show that the best model is achieved with a learning rate of 1e-3, batch size of 64, and 150 epochs, achieving a performance of mAP@50 of 0.962 and mAP@50–95 of 0.746 on the validation data. On the test data, the model achieves a precision of 0.828, recall of 0.783, mAP@50 of 0.836, and mAP@50–95 of 0.628, indicating good generalization capability. Furthermore, the application of image enhancement significantly improves detection performance, and the resulting model has low computational complexity and fast inference time, making it suitable for real-time implementation based on edge computing to support quality control processes in the mining industry.

---

Availability

Detail Information

Series Title

-

Call Number

T2004232026

Publisher

Indralaya : Prodi Sistem Komputer, Fakultas Ilmu Komputer Universitas Sriwijaya., 2026

Collation

xiv, 81 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

621.867 207

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Sistem Komputer
Sistem Konveyor

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment