Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Format: | Recurso digital |
| Sprache: | |
| Veröffentlicht: |
Zenodo
2025
|
| Online-Zugang: | https://doi.org/10.5281/zenodo.17583551 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Inhaltsangabe:
- <p><strong>Abstract</strong><br>Structural Health Monitoring (SHM) is essential for ensuring the safety and reliability of critical assets across civil, aerospace, and industrial domains. Traditional inspection techniques rely heavily on manual assessments, which are time-consuming, subjective, and prone to oversight. To address these limitations, this work presents an automated computer-vision–based SHM system for crack and corrosion detection in structural surfaces.</p> <p>The proposed framework integrates a <strong>YOLOv8 deep learning model</strong> for high-accuracy defect localization with the <strong>Simple Online Realtime Tracking (SORT)</strong> algorithm to maintain persistent identities of defects across video frames. A customized severity-estimation module computes geometric indicators, such as defect length and area, to quantify progression. The system outputs annotated videos, per-frame logs of defect metrics, and automatically generated PDF reports summarizing defect evolution. In addition, a <strong>Streamlit-based user interface</strong> enables simplified end-to-end execution, including video upload, processing, visualization, and report export.</p> <p>Experimental evaluation demonstrates that the integrated YOLOv8–SORT pipeline delivers efficient and reliable performance for automated crack and corrosion detection. The framework provides a scalable, real-time decision-support tool for structural inspection and maintenance planning.</p> <p><strong>Keywords:</strong> Structural Health Monitoring, YOLOv8, SORT, Crack Detection, Corrosion Detection, Computer Vision, Deep Learning, Predictive Maintenance</p> <p><strong>Source Code:</strong> <a href="https://github.com/1225zisu-lab/Structural-Health-Monitoring" target="_new" rel="noopener">https://github.com/1225zisu-lab/Structural-Health-Monitoring</a><br><strong>DOI:</strong> 10.5281/zenodo.17583551</p> <p>This work builds upon prior independent research on defect detection for metallic surfaces, now unified within a single SHM framework.</p>