If you have any problems related to the accessibility of any content (or if you want to request that a specific publication be accessible), please contact us at firstname.lastname@example.org.
Deep Learning Based Concrete Distress Detection System for Civil Infrastructure
AdvisorLa, Hung (Jim)
AltmetricsView Usage Statistics
In most civil concrete structures, the inspection of structural health is essential. A periodical inspection process ensures that the infrastructure will meet the functional requirements properly or not. To avoid hazardous situations in civil infrastructure, proper maintenance of concrete structures is necessary. The manual visual examination process might provide erroneous results while exploring critical parts of concrete surfaces. As a result, an accurate, safe, and dependable automated process is required for detecting concrete distress. Spalling is a critical distress that can damage concrete surfaces in civil infrastructure. Severe and harmful spalling needs to be taken care of to avoid life-threatening incidents by identifying the location of the distress. Aside from determining the location of the spalling, the severity level of the spalling must also be determined. These severity levels help determine how adverse the situation is and prioritize the process of fixing the spalling. Due to the impact of concrete distress, detecting surface defects like spallings caught the attention of researchers. In this thesis, we have presented approaches to detecting the location of spalling according to its severity level. The proposed methods use deep learning-based approaches and multi-class semantic segmentation. Our approaches have explored two major criteria to detect the spalling and categorize its severity level. Furthermore, we have conducted qualitative and quantitative analyses to demonstrate the performance achieved by the proposed methodologies.