“In the field of crack detection on building exterior walls, researchers have proposed a crack detection method based on aerial images and computer vision. By collecting images through drones and optimizing the U-Net model, the efficiency, accuracy, and safety of crack detection have been effectively improved. The improved U-Net model has improved by 3.53% and 4.18% in IoU metrics and F1_store, respectively, outperforming classical segmentation models, providing a safe, efficient, and accurate solution for detecting cracks on building exterior walls.”