Leveraging Segment Anything Model (SAM) for Weld Defect Detection in Industrial Ultrasonic B-Scan Images
Sensors · 06.01.2025 · 25 (1) · 277
DOI: 10.3390/s25010277 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, online erschienen am 06.01.2025. Verfasst von Amir-M Naddaf-Sh und 2 weiteren Autor:innen. Beteiligte Einrichtung: Phillip M. Drayer Electrical Engineering Department, Lamar University, Beaumont, TX 77705, USA. Die Arbeit wurde bisher 15-mal zitiert (Crossref) und nennt 90 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Lamar University's Center for Midstream Management and Science (CMMS), Lamar University’s Center for Midstream Management and Science (CMMS).
Abstract (Originalsprache)
Automated ultrasonic testing (AUT) is a critical tool for infrastructure evaluation in industries such as oil and gas, and, while skilled operators manually analyze complex AUT data, artificial intelligence (AI)-based methods show promise for automating interpretation. However, improving the reliability and effectiveness of these methods remains a significant challenge. This study employs the Segment Anything Model (SAM), a vision foundation model, to design an AI-assisted tool for weld defect detection in real-world ultrasonic B-scan images. It utilizes a proprietary dataset of B-scan images generated from AUT data collected during automated girth weld inspections of oil and gas pipelines, detecting a specific defect type: lack of fusion (LOF). The implementation includes integrating knowledge from the B-scan image context into the natural image-based SAM 1 and SAM 2 through a fully automated, promptable process. As part of designing a practical AI-assistant tool, the experiments involve applying both vanilla and low-rank adaptation (LoRA) fine-tuning techniques to the image encoder and mask decoder of different variants of both models, while keeping the prompt encoder unchanged. The results demonstrate that the utilized method achieves improved performance compared to a previous study on the same dataset.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Phillip M. Drayer Electrical Engineering Department, Lamar University, Beaumont, TX 77705, USA.
- CRC-Evans, Houston, TX 77066, USA.
ORCID-Profile: Amir-M Naddaf-Sh, Hassan Zargarzadeh
Förderung
- Lamar University's Center for Midstream Management and Science (CMMS)
- Lamar University’s Center for Midstream Management and Science (CMMS)
Zitieren
APA
Naddaf-Sh, A. M., Baburao, V. S., & Zargarzadeh, H. (2025). Leveraging Segment Anything Model (SAM) for Weld Defect Detection in Industrial Ultrasonic B-Scan Images. Sensors, 25(1), 277. https://doi.org/10.3390/s25010277
Vancouver
Naddaf-Sh AM, Baburao VS, Zargarzadeh H. Leveraging Segment Anything Model (SAM) for Weld Defect Detection in Industrial Ultrasonic B-Scan Images. Sensors. 2025;25(1):277. doi:10.3390/s25010277
BibTeX
@article{NaddafSh2025,
author = {Naddaf-Sh, Amir-M and Baburao, Vinay S and Zargarzadeh, Hassan},
title = {Leveraging Segment Anything Model (SAM) for Weld Defect Detection in Industrial Ultrasonic B-Scan Images},
journal = {Sensors},
year = {2025},
volume = {25},
number = {1},
pages = {277},
doi = {10.3390/s25010277}
}
EUVIMED