Speaker
Description
Over time, buildings are increasingly challenged by various defects, such as cracks, mold growth, paint deterioration, and dampness. Addressing these defects is crucial to prevent excessive maintenance costs and potential disasters, but effective treatment relies on accurately identifying and detecting these issues. Traditionally, engineers have relied on building surveys to locate and assess these defects. However, manual inspections are labor-intensive, time-consuming, and often require skilled labor, increasing costs. Moreover, the accuracy of results from manual methods may be inconsistent, and accessing defects in hard-to-reach or hazardous areas poses significant challenges and risks, sometimes making detection infeasible. To overcome these limitations, this study proposes an application of an Image processing and Convolutional Neural Network (CNN) approach for defect detection using images captured by regular portable cell phone cameras. The method employs the pre-trained Convolutional Neural Network (CNN) model to detect building defects. The model was trained, validated, and tested on a dataset of images collected from various buildings. The dataset was categorized into cracks, non-cracks, and miscellaneous defects. The comparative study exhibited that the results of the XXX model are the best among the used model.
