🤖 AI Summary
This work addresses the limited geographic coverage of existing building change detection datasets, which hinders the evaluation of model generalization across regions. To this end, we construct a large-scale benchmark dataset spanning 28 French departments, comprising approximately 28,000 pairs of 0.2-meter-resolution bitemporal remote sensing images with corresponding pixel-level change masks. Geographic isolation is enforced between training, validation, and test sets to enable rigorous assessment of domain transfer. Derived from IGN orthophotos and building footprints, the dataset includes both pixel- and instance-level annotations, enhanced by manual verification to ensure high quality and complete spatial metadata. Experiments demonstrate that models trained on this geographically diverse benchmark significantly outperform those trained on existing datasets such as LEVIR-CD+ and WHU-CD in cross-regional scenarios, underscoring the critical role of geographic diversity in improving model generalization.
📝 Abstract
We introduce FOTBCD, a large-scale building change detection dataset derived from authoritative French orthophotos and topographic building data provided by IGN France. Unlike existing benchmarks that are geographically constrained to single cities or limited regions, FOTBCD spans 28 departments across mainland France, with 25 used for training and three geographically disjoint departments held out for evaluation. The dataset covers diverse urban, suburban, and rural environments at 0.2m/pixel resolution. We publicly release FOTBCD-Binary, a dataset comprising approximately 28,000 before/after image pairs with pixel-wise binary building change masks, each associated with patch-level spatial metadata. The dataset is designed for large-scale benchmarking and evaluation under geographic domain shift, with validation and test samples drawn from held-out departments and manually verified to ensure label quality. In addition, we publicly release FOTBCD-Instances, a publicly available instance-level annotated subset comprising several thousand image pairs, which illustrates the complete annotation schema used in the full instance-level version of FOTBCD. Using a fixed reference baseline, we benchmark FOTBCD-Binary against LEVIR-CD+ and WHU-CD, providing strong empirical evidence that geographic diversity at the dataset level is associated with improved cross-domain generalization in building change detection.