Klasifikasi Pola Retak Beton Pada Permukaan Bangunan Secara Real-time Menggunakan You Only Look Once

Herfieda, Dewi Cantika (2024) Klasifikasi Pola Retak Beton Pada Permukaan Bangunan Secara Real-time Menggunakan You Only Look Once. Diploma thesis, Institut Teknologi Sepuluh Nopember.

Text 2040201083-Undergraduate_Thesis for D4 program.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (6MB) | Request a copy

Abstract

Penelitian ini mengembangkan dan mengevaluasi model klasifikasi pola retak beton pada permukaan bangunan secara real-time menggunakan metode You Only Look Once (YOLO). Tujuan penelitian ini adalah untuk mengotomatiskan proses inspeksi visual guna mengidentifikasi kerusakan bangunan. Hal tersebut mendukung strategi pemeliharaan efisien, dan pemantauan kesehatan struktural. Tahap pemantauan kesehatan bangunan dilakukan dengan mengamati kerusakan seperti retakan, lendutan, atau korosi untuk pengujian lanjutan. Dua model utama yang diuji adalah YOLOv5x dan YOLOv8s. Hasil penelitian menunjukkan bahwa YOLOv8s lebih unggul dalam kecepatan dengan frame rate 4 fps, sedangkan YOLOv5x menunjukkan performa serupa dalam berbagai metrik evaluasi. Pada hasil pengujian YOLOv5x mencapai presisi 85,5% pada Gambar, 83,4% dengan pada Video, dengan Recall 81,2% 85,5% pada Gambar, 77% pada Video serta Akurasi 97,1%. pada Gambar, 96,5% pada Video. Sedangkan, YOLOv8s mencapai Pecision 90% pada Gambar, 78,2% pada Video dengan Recall 87,5% pada Gambar, 78,3% pada Video serta Akurasi 98,3% pada Gambar, 96,3% pada Video. Selain itu, fitur yang berhasil dikembangkan dari model ini berupa website aplikasi meliputi pelacakan objek, pengumpulan data jumlah kerusakan retak, dan pengambilan gambar retak.
=================================================================================================================================
This research develops and evaluates a real-time classification model of concrete crack patterns on building surfaces using the You Only Look Once (YOLO) method. The aim of this research is to automate the visual inspection process to identify building defects. This supports efficient maintenance strategies, and structural health monitoring. The health monitoring phase of the building is done by observing damage such as cracks, deflection, or corrosion for further testing. The two main models tested were YOLOv5x and YOLOv8s. The results showed that YOLOv8s was superior in speed with a frame rate of 4 fps, while YOLOv5x showed similar performance in various evaluation metrics. In the test results, YOLOv5x achieved 85,5% precision on image, 83,4% with on video, with 81,2% recall 85,5% on image, 77% on video and 97.1% accuracy on Image, 96.5% on video. Meanwhile, YOLOv8s achieved 90% precision in image, 78,2% in video with 87,5% recall in image, 78,3% in video and 98,3% accuracy in image, 96,3% in video. In addition, the features that were successfully developed from this model in the form of an application website include object tracking, data collection on the amount of crack damage, and crack image capture.

Item Type:	Thesis (Diploma)
Uncontrolled Keywords:	Pemantauan Kesehatan Struktural, Retak Beton, Klasifikasi Retak, You Only Look Once (YOLO), Structural Health Monitoring, Concrete Cracking, Crack Classification
Subjects:	T Technology > T Technology (General) T Technology > T Technology (General) > T11 Technical writing. Scientific Writing T Technology > T Technology (General) > T174 Technological forecasting T Technology > T Technology (General) > T58.62 Decision support systems T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques T Technology > TA Engineering (General). Civil engineering (General) > TA440 Concrete--Cracking.
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Dewi Cantika Herfieda
Date Deposited:	16 Aug 2024 08:24
Last Modified:	16 Aug 2024 08:24
URI:	http://repository.its.ac.id/id/eprint/115448

Actions (login required)

View Item