Consider Computer Vision for Building Component Recognition and Defect Detection

With the advancement of computer vision technology, its application in the construction field has become increasingly widespread, particularly in the identification of building structural components and defect detection. This paper investigates the use of computer vision techniques for the recognition and defect detection of building structural components. We employed the DeeplabV3+ model, which is based on MobileNetV2 and utilizes an encoder-decoder architecture along with a pyramid pooling module for semantic segmentation. The experiments were conducted on the Windows 10 operating system using the Python programming language and the TensorFlow framework, with meticulous parameter settings for model training. Model performance was evaluated using metrics such as Mean Pixel Accuracy (MPA), Mean Intersection over Union (MIoU), and F1 score. The results demonstrated satisfactory performance on both the training and validation sets, with F1 scores of 0.7180 and 0.6520, respectively, thereby validating its effectiveness in the recognition of building structural components.