Skripsi

DETEKSI DAN ESTIMASI DIMENSI LUBANG JALAN SECARA REAL-TIME BERBASIS DEEP LEARNING

---

Real-time pothole detection and dimension estimation are crucial aspects of various applications, such as road maintenance and driving safety. However, previous studies only focused on detection, or dimension estimation, or a combination of both, but not in real-time. Therefore, in this study, the detection, dimension estimation, and distance estimation are combined in real-time. The dataset consists of various shapes and sizes of potholes taken from roads in the Province of Sumatra Selatan and the City of Palembang. The YOLOv8x-seg model was trained for 50, 100, and 200 epochs to find the optimal configuration. The best result was obtained with 50 epochs, achieving a box mean average precision (mAP) of 0.816 and a box loss of 0.659. Simulation testing showed that the developed model is capable of carrying out detection, dimension estimation, and distance estimation tasks with average errors for width, height, and distance of 13.88%, 19.85%, and 24.55%. Furthermore, real-time implementation showed that the model could perform all the tasks with average errors for width, height, and distance of 11.44%, 15.57%, and 12.24%. These results indicate that the developed model could be effectively used for road monitoring and maintenance, as well as reducing potential hazards around detected potholes, particularly in the Province of Sumatra Selatan and the City of Palembang.

---

Availability

Detail Information

Series Title

-

Call Number

T1515092024

Publisher

Indralaya : Prodi Teknik Elektro, Fakultas Teknik Universitas Sriwijaya., 2024

Collation

xiii, 35 hlm.; ilus.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

006.370 7

Content Type

Text

Media Type

unmediated

Carrier Type

-

Edition

-

Subject(s)

Prodi Teknik Elektro

Specific Detail Info

-

Statement of Responsibility

KA

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment