[Paper Review] Application of Segment Anything Model for Civil Infrastructure Defect Assessment

Mohsen Ahmadi, Ahmad Gholizadeh Lonbar+5 more|arXiv (Cornell University)|Apr 25, 2023

Infrastructure Maintenance and Monitoring23 citations

TL;DR

The paper compares SAM and U-Net for crack detection in concrete, showing complementary strengths and proposing a combined approach to improve civil infrastructure defect assessment. It discusses the potential impact on bridges, buildings, and roads.

ABSTRACT

This research assesses the performance of two deep learning models, SAM and U-Net, for detecting cracks in concrete structures. The results indicate that each model has its own strengths and limitations for detecting different types of cracks. Using the SAM's unique crack detection approach, the image is divided into various parts that identify the location of the crack, making it more effective at detecting longitudinal cracks. On the other hand, the U-Net model can identify positive label pixels to accurately detect the size and location of spalling cracks. By combining both models, more accurate and comprehensive crack detection results can be achieved. The importance of using advanced technologies for crack detection in ensuring the safety and longevity of concrete structures cannot be overstated. This research can have significant implications for civil engineering, as the SAM and U-Net model can be used for a variety of concrete structures, including bridges, buildings, and roads, improving the accuracy and efficiency of crack detection and saving time and resources in maintenance and repair. In conclusion, the SAM and U-Net model presented in this study offer promising solutions for detecting cracks in concrete structures and leveraging the strengths of both models that can lead to more accurate and comprehensive results.

Motivation & Objective

• Motivate improved crack detection in concrete structures using advanced DL methods.
• Evaluate the performance of Segment Anything Model (SAM) for crack localization.
• Evaluate the performance of U-Net for determining crack size and location.
• Explore how combining SAM and U-Net can yield more accurate defect assessments.
• Highlight potential applications for bridges, buildings, and roads in maintenance planning.

Proposed method

• Compare SAM and U-Net on concrete crack datasets to identify strengths and limitations.
• Analyze SAM’s image-part segmentation approach for locating cracks, especially longitudinal cracks.
• Analyze U-Net’s ability to identify positive label pixels for crack size and location.
• Propose a combined workflow leveraging both models for enhanced defect detection.
• Discuss implications of using SAM and U-Net for various concrete structures.

Experimental results

Research questions

• RQ1 How do SAM and U-Net perform in detecting different types of cracks in concrete structures?
• RQ2 What are the strengths and limitations of SAM in crack localization versus U-Net in crack sizing?
• RQ3 Can combining SAM and U-Net improve overall crack detection accuracy and coverage?
• RQ4 What are the practical implications for civil infrastructure maintenance and safety?

Key findings

• SAM shows strengths in locating cracks through its partition-based approach, particularly for longitudinal cracks.
• U-Net excels at identifying positive label pixels to accurately detect crack size and location.
• Using SAM and U-Net together yields more accurate and comprehensive crack detection results than either model alone.
• Both models offer promising capabilities for detecting cracks in various concrete structures such as bridges, buildings, and roads.
• The study emphasizes the importance of advanced technologies for improving safety and longevity of concrete infrastructure.