UrbanCDNet: Appearance-Robust and Boundary-Aware Bitemporal Change Detection for Korean Urban Building Monitoring
2026-06-29 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors developed UrbanCDNet, a special deep learning model that helps detect changes in buildings using aerial images taken at different times in Korean cities. Their model is designed to handle tricky situations like small changed areas and big differences in image appearance between dates. They tested it on a large Korean dataset and found it works better than other strong models, especially when changes are sparse or images look very different. The results show that focusing on how changes are compared over time and paying attention to building boundaries improves detection more than just using bigger models.
Urban building change detectionBi-temporal aerial imagerySiamese CNNMulti-cue comparisonBoundary supervisionF1 scoreIntersection over Union (IoU)Sparse change detectionPhotometric gapSiamese U-Net
Authors
Abdirashid Omar, Jonghyuk Park
Abstract
Urban building change detection from bi-temporal aerial imagery is important for redevelopment monitoring, infrastructure management, and unauthorized-construction screening, but Korean urban scenes remain difficult because changed regions are often sparse, appearance varies strongly between acquisition dates, and useful outputs must follow building footprints rather than coarse blobs. This paper presents UrbanCDNet, a task specific Siamese CNN that combines appearance-robust multi-cue comparison, alignment-aware middle-scale differencing, lightweight context refinement, scene calibration, and auxiliary boundary supervision. Experiments use a corrected AIHub-based Korean benchmark with 3,998 training, 503 validation, and 499 test pairs, and report changed-class precision, recall, F1, and IoU. On the locked test split, UrbanCDNet achieves 0.7335 precision, 0.7696 recall, 0.7511 F1, and 0.6014 IoU, outperforming a strong Siamese U-Net baseline (0.7108 F1, 0.5514 IoU) and the strongest external competitor, ChangeFormer-MIT-B0 (0.7107 F1, 0.5512 IoU). Additional diagnostic slicing shows that the gain is concentrated in the operating regimes that motivated the design: on the sparse-change subset with less than 5% changed area, F1 improves from 0.4765 to 0.6175, and on the high photometric-gap subset it improves from 0.6349 to 0.7285. Boundary F1 at 3-pixel tolerance rises from 0.3445 to 0.4447, while object F1 at IoU 0.3 rises from 0.0690 to 0.2258. These results indicate that, on this Korean benchmark, task-shaped temporal comparison and boundary-aware supervision matter more than generic model scale alone