Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Department of Building and Real Estate | en_US |
| dc.contributor.advisor | Kim, Minkoo (BRE) | en_US |
| dc.creator | Huang, Yueming | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/10747 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | en_US |
| dc.rights | All rights reserved | en_US |
| dc.title | Deep-learning-based detection for steel coating surface defects identification and localization using images | en_US |
| dcterms.abstract | Coatings are widely used in steel structural components to improve the fire resistance and corrosion resistance of steel components. However, the damage of the coating will reduce the fire resistance and anti-corrosion performance of the steel structure, and even cause the steel structure members to directly lose their strength and cause structural damage. Therefore, the detection of steel structure surface damage is particularly important in the process of structural maintenance. Current monitoring of steel structure surface damage relies heavily on manual inspection. However, steel structural members that are likely to cause structural damage are usually located in areas such as roofs that are difficult to perform manual inspection, which making the manual inspection process difficult and time-consuming, and for some components that are difficult to perform manual inspection, manual inspection is difficult to carry out. This dissertation proposes a deep learning-based steel structure surface damage detection method using images, which can use the images to classify and localize the damage of the steel structure surface simultaneously. A defect classifier based on CNN's VGG-16 with accuracy 98.31% is developed to automatically diagnose what type of defects of the investigated surface region, and use the trained CNN model with Grad-CAM algorithm to explain the results and localizing the defects regions. In addition, 258 images collected on-site in Hong Kong is to verify the effectiveness and accuracy of the proposed defect detection method. Furthermore, a parametric study with varying deep learning paraments is performed for optimal classification accuracy and efficiently convergence. The results demonstrate that the proposed method can properly estimate the location and class of the steel components surface defects. | en_US |
| dcterms.extent | xi, 93 pages : color illustrations | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 2020 | en_US |
| dcterms.educationalLevel | M.Sc. | en_US |
| dcterms.educationalLevel | All Master | en_US |
| dcterms.LCSH | Steel, Structural | en_US |
| dcterms.LCSH | Protective coatings | en_US |
| dcterms.LCSH | Steel -- Surfaces --Defects | en_US |
| dcterms.LCSH | Corrosion and anti-corrosives | en_US |
| dcterms.LCSH | Hong Kong Polytechnic University -- Dissertations | en_US |
| dcterms.accessRights | restricted access | en_US |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 5190.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 5.37 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/200/10747