Expand operator samples for YOLO hard negatives

scripts/prepare_operator_real_data_samples.py

@@ -36,6 +36,8 @@ class OperatorSample:
     half_size_m: float = 250.0
     width: int = 512
     height: int = 512
+    sample_role: str = "reference"
+    allow_empty_reference: bool = False
 
 
 SAMPLES: dict[str, OperatorSample] = {
@@ -57,6 +59,61 @@ SAMPLES: dict[str, OperatorSample] = {
         center_lon=4.9488,
         center_lat=51.3225,
     ),
+    "herentals": OperatorSample(
+        slug="herentals",
+        display_name="Herentals center",
+        center_lon=4.8339,
+        center_lat=51.1766,
+        half_size_m=220.0,
+    ),
+    "balen": OperatorSample(
+        slug="balen",
+        display_name="Balen center",
+        center_lon=5.1703,
+        center_lat=51.1688,
+        half_size_m=220.0,
+    ),
+    "retie": OperatorSample(
+        slug="retie",
+        display_name="Retie center",
+        center_lon=5.0827,
+        center_lat=51.2665,
+        half_size_m=220.0,
+    ),
+    "westerlo": OperatorSample(
+        slug="westerlo",
+        display_name="Westerlo center",
+        center_lon=4.9158,
+        center_lat=51.0909,
+        half_size_m=220.0,
+    ),
+    "postel_bos": OperatorSample(
+        slug="postel_bos",
+        display_name="Postel forest background candidate",
+        center_lon=5.16,
+        center_lat=51.305,
+        half_size_m=260.0,
+        sample_role="background_candidate",
+        allow_empty_reference=True,
+    ),
+    "lommel_heide": OperatorSample(
+        slug="lommel_heide",
+        display_name="Lommel forest background candidate",
+        center_lon=5.287,
+        center_lat=51.249,
+        half_size_m=260.0,
+        sample_role="background_candidate",
+        allow_empty_reference=True,
+    ),
+    "kasterlee_bos": OperatorSample(
+        slug="kasterlee_bos",
+        display_name="Kasterlee forest background candidate",
+        center_lon=4.965,
+        center_lat=51.273,
+        half_size_m=260.0,
+        sample_role="background_candidate",
+        allow_empty_reference=True,
+    ),
 }
 
 
@@ -218,7 +275,7 @@ def fetch_reference(sample: OperatorSample, reference_path: Path, geo_bbox: list
     response.raise_for_status()
     reference = response.json()
     features = reference.get("features") or []
-    if not features:
+    if not features and not sample.allow_empty_reference:
         raise SystemExit(f"GRB GBG returned no building features for {sample.slug} bbox {geo_bbox}")
 
     reference["name"] = f"GRB GBG buildings - {sample.display_name} sample AOI"
@@ -227,11 +284,14 @@ def fetch_reference(sample: OperatorSample, reference_path: Path, geo_bbox: list
     reference["attribution"] = "Bron: Grootschalig Referentie Bestand Vlaanderen, Digitaal Vlaanderen"
     reference["bbox"] = geo_bbox
     reference["sample_slug"] = sample.slug
+    reference["sample_role"] = sample.sample_role
+    reference["allow_empty_reference"] = sample.allow_empty_reference
     for feature in features:
         props = feature.setdefault("properties", {})
         props.setdefault("source_name", "grb")
         props.setdefault("reference_layer_name", "buildings")
         props.setdefault("sample_slug", sample.slug)
+        props.setdefault("sample_role", sample.sample_role)
 
     reference_path.write_text(json.dumps(reference, ensure_ascii=False), encoding="utf-8")
     return prepared_url(GRB_GBG_URL, ogc_params), len(features)
@@ -256,6 +316,8 @@ def prepare_sample(sample: OperatorSample, output_dir: Path, force: bool) -> dic
         "center_lon": sample.center_lon,
         "center_lat": sample.center_lat,
         "half_size_m": sample.half_size_m,
+        "sample_role": sample.sample_role,
+        "allow_empty_reference": sample.allow_empty_reference,
         "raster_path": str(ortho_path),
         "reference_path": str(reference_path),
         "reference_feature_count": reference_feature_count,
@@ -288,7 +350,7 @@ def write_readme(output_dir: Path, samples: list[dict[str, Any]]) -> None:
         lines.append(
             f"- `{sample['sample_slug']}`: `{Path(sample['raster_path']).name}` and "
             f"`{Path(sample['reference_path']).name}`, "
-            f"{sample['reference_feature_count']} reference features."
+            f"{sample['reference_feature_count']} reference features, role `{sample['sample_role']}`."
         )
     lines.append("")
     lines.append("Purpose: configured-YOLO detection + persisted QA/QC validation with operator-provided files.")