Add dated regional imagery training sources

scripts/assemble_belgium_building_corpus.py

@@ -28,9 +28,9 @@ from app.models import Dataset  # noqa: E402
 from normalize_belgium_building_labels import normalize  # noqa: E402
 
 REGION_SOURCES = {
-    "flanders": ("digitaal_vlaanderen_orthophoto", "grb"),
-    "wallonia": ("spw_orthophoto", "spw_picc"),
-    "brussels": ("urbis_orthophoto", "urbis"),
+    "flanders": ({"digitaal_vlaanderen_orthophoto"}, "grb"),
+    "wallonia": ({"spw_orthophoto"}, "spw_picc"),
+    "brussels": ({"urbis_orthophoto", "digitaal_vlaanderen_orthophoto"}, "urbis"),
 }
 SPLITS = {"train", "val", "calibration", "test", "background-test"}
 
@@ -56,8 +56,8 @@ def _validate_pair(sample: dict[str, Any], raster: Dataset, reference: Dataset)
     region = str(sample.get("region") or "").lower()
     if region not in REGION_SOURCES:
         raise SystemExit(f"Unsupported region for {sample.get('sample_slug')}: {region}")
-    expected_raster, expected_reference = REGION_SOURCES[region]
-    if raster.source_name != expected_raster:
+    expected_rasters, expected_reference = REGION_SOURCES[region]
+    if raster.source_name not in expected_rasters:
         raise SystemExit(f"Raster provider mismatch for {sample['sample_slug']}: {raster.source_name}")
     if reference.source_name != expected_reference or reference.reference_layer_name != "buildings":
         raise SystemExit(f"Reference provider/layer mismatch for {sample['sample_slug']}")
@@ -156,6 +156,7 @@ def main() -> int:
                 {
                     "sample_slug": slug,
                     "sample_role": sample.get("sample_role", "positive"),
+                    "require_empty": bool(sample.get("require_empty", False)),
                     "region": region,
                     "context": sample.get("context"),
                     "split": sample["split"],