Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Capturing and analysis of facade images of historical buildings to detect defects in order to assess their damage risk

Koutmos Vasileios

Full record

---

Summary

The exterior appearance and structural stability of buildings are negatively impacted by defects in the façades of residential and historical structures. During maintenance, manual labor is usually used to address façade defects in buildings. This method takes a long time, produces arbitrary outcomes, and may end in mishaps or casualties. Ultimately, it is ideal to prevent all types of defects in the design or construction stages, but this is a very difficult goal to achieve. Thus, there is a need for a method to effectively monitor defects in the maintenance phase and actively respond to the occurrence of the defects. Therefore, it is necessary to develop a technology that can continually and automatically monitor defects in residential buildings that minimize the dependence on manpower. Furthermore, there are various types of defects in residential building, and each defect type in the real world appears in an irregular pattern. To consider the characteristics of these defects, automated defect monitoring technology should be able to simultaneously detect and effectively classify various types of defects in image data.To address this proposal, plenty of methods have been implemented, utilizing different types of deep learning models. The current thesis emphasizes on two different methodologies, which are expanded upon and combined, in order to more efficiently manage defects by minimizing the involvement of manpower.The dataset used for training a deep-learning-based network contains actual residential and historical building façade images. Faster regions with a convolutional neural network (Faster R-CNN) structure are employed for more accurate defect detection in such environments. As it is difficult to detect defects in a training environment, it is necessary to improve the performance of the network. However, the object detection network employed in this dissertation yields an excellent performance in complex real-world images, indicating the possibility of developing a system that would detect a great number of defects in more types of building façades.Summarizing the contents of the present thesis, that combines two distinct methodologies, and presents the results from the implementation on real historical buildings. The ultimate purpose of the paper is to expand the use of non-traditional methods in defects in historical building’s façades, that depend less on manpower. In the end of the dissertation, the results are going to be evaluated in terms of their accuracy and their efficiency, concluding in some key aspects the field will benefit from the transition to a more automated model.

Available Files

Services