| Author: | Xiong, Chenqin |
| Title: | AI-based models for detecting and quantifying surface defects of concrete bridge decks |
| Advisors: | Zayed, Tarek (BRE) |
| Degree: | M.Sc. |
| Year: | 2023 |
| Subject: | Bridges -- Maintenance and repair Structural health monitoring Hong Kong Polytechnic University -- Dissertations |
| Department: | Department of Building and Real Estate |
| Pages: | ix, 68 pages : color illustrations |
| Language: | English |
| Abstract: | This dissertation presents a comprehensive investigation into bridge crack recognition, detection, and measurement, focusing on the Hong Kong context. The study encompasses three key findings that significantly contribute to advancements in this field. Firstly, a new dataset comprising images of bridge cracks specific to Hong Kong was successfully constructed. This dataset was utilized for training Convolutional Neural Networks (CNNs) models, resulting in improved robustness and generalization capabilities. This dataset's introduction addresses existing datasets limitations and facilitates more accurate and reliable crack detection. Secondly, an enhanced version of the YOLOv8 model was developed, explicitly tailored for bridge crack detection. This model outperformed other popular models in terms of accuracy and precision. Evaluation of the validation set demonstrated promising results, with a Mean Average Precision (Map50) of 95.90% and Map50:95 of 71.10%. These findings underscore the effectiveness of the tuned YOLOv8 model in accurately detecting bridge cracks, offering significant improvements over existing approaches. Lastly, the study explored crack measurement models, which exhibited strong performance when applied to images depicting single cracks. These models successfully measured the geometric parameters of segmented crack images, providing reliable and accurate measurements for practical applications in bridge inspection. This aspect contributes to the practical applicability of the study's findings, enabling more precise evaluation and assessment of bridge cracks. Overall, this dissertation's key findings highlight significant contributions in terms of dataset creation, model development, and measurement capabilities. The successful construction of a new dataset specific to the Hong Kong context and the improved performance of the YOLOv8 model demonstrate notable advancements in bridge crack detection. The effective crack measurement models also enhance the practical applicability of the study's findings, contributing to improved bridge crack recognition, detection, and measurement techniques. Ultimately, these advancements enhance the efficiency and accuracy of bridge inspection processes, benefiting infrastructure maintenance and safety initiatives. |
| Rights: | All rights reserved |
| Access: | restricted access |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 7078.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 2.97 MB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/12614