Expand YOLO training AOIs safely

scripts/prepare_operator_real_data_samples.py

@@ -25,6 +25,12 @@ DEFAULT_GRB_MAX_FEATURES = 100000
 REFERENCE_AOI_CATEGORY = "reference_aoi"
 PURE_EMPTY_BACKGROUND_CATEGORY = "pure_empty_negative"
 SPARSE_BACKGROUND_CATEGORY = "sparse_building_context"
+TRAINING_EXPANSION_SAMPLE_SLUGS = frozenset(
+    {"olen_center", "lille_center", "oud_turnhout_center", "kasterlee_center"}
+)
+DEFAULT_VALIDATION_SAMPLE_SLUGS = frozenset(
+    {"turnhout", "retie", "westerlo", "arendonk_heide"}
+)
 requests: Any = None
 rasterio: Any = None
 Transformer: Any = None
@@ -92,6 +98,30 @@ SAMPLES: dict[str, OperatorSample] = {
         center_lat=51.0909,
         half_size_m=220.0,
     ),
+    "olen_center": OperatorSample(
+        slug="olen_center",
+        display_name="Olen center training expansion",
+        center_lon=4.8597257,
+        center_lat=51.1438611,
+    ),
+    "lille_center": OperatorSample(
+        slug="lille_center",
+        display_name="Lille center training expansion",
+        center_lon=4.8242404,
+        center_lat=51.2382180,
+    ),
+    "oud_turnhout_center": OperatorSample(
+        slug="oud_turnhout_center",
+        display_name="Oud-Turnhout center training expansion",
+        center_lon=4.9817086,
+        center_lat=51.3178319,
+    ),
+    "kasterlee_center": OperatorSample(
+        slug="kasterlee_center",
+        display_name="Kasterlee center training expansion",
+        center_lon=4.9678120,
+        center_lat=51.2407915,
+    ),
     "postel_bos": OperatorSample(
         slug="postel_bos",
         display_name="Postel forest background candidate",
@@ -270,6 +300,10 @@ def background_category_for_sample(sample: OperatorSample, reference_feature_cou
     return PURE_EMPTY_BACKGROUND_CATEGORY if reference_feature_count <= 0 else SPARSE_BACKGROUND_CATEGORY
 
 
+def recommended_split_for_sample(sample: OperatorSample) -> str:
+    return "val" if sample.slug in DEFAULT_VALIDATION_SAMPLE_SLUGS else "train"
+
+
 def apply_sample_overrides(
     sample: OperatorSample,
     *,
@@ -481,6 +515,7 @@ def fetch_reference(
     reference["sample_role"] = sample.sample_role
     reference["allow_empty_reference"] = sample.allow_empty_reference
     reference["background_category"] = background_category_for_sample(sample, len(features))
+    reference["recommended_split"] = recommended_split_for_sample(sample)
     reference["reference_page_limit"] = page_limit
     reference["reference_max_features"] = max_features
     reference["reference_pages_fetched"] = len(pages)
@@ -495,6 +530,7 @@ def fetch_reference(
         props.setdefault("sample_slug", sample.slug)
         props.setdefault("sample_role", sample.sample_role)
         props.setdefault("background_category", background_category_for_sample(sample, len(features)))
+        props.setdefault("recommended_split", recommended_split_for_sample(sample))
 
     reference_path.write_text(json.dumps(reference, ensure_ascii=False), encoding="utf-8")
     return prepared_url(GRB_GBG_URL, ogc_params), len(features)
@@ -537,6 +573,7 @@ def prepare_sample(
         "sample_role": sample.sample_role,
         "allow_empty_reference": sample.allow_empty_reference,
         "background_category": background_category,
+        "recommended_split": recommended_split_for_sample(sample),
         "raster_path": str(ortho_path),
         "reference_path": str(reference_path),
         "reference_feature_count": reference_feature_count,
@@ -572,7 +609,8 @@ def write_readme(output_dir: Path, samples: list[dict[str, Any]]) -> None:
             f"- `{sample['sample_slug']}`: `{Path(sample['raster_path']).name}` and "
             f"`{Path(sample['reference_path']).name}`, "
             f"{sample['reference_feature_count']} reference features, role `{sample['sample_role']}`, "
-            f"background category `{sample['background_category']}`."
+            f"background category `{sample['background_category']}`, "
+            f"recommended split `{sample['recommended_split']}`."
         )
     lines.append("")
     lines.append("Purpose: configured-YOLO detection + persisted QA/QC validation with operator-provided files.")
@@ -602,7 +640,7 @@ def main() -> int:
     write_readme(output_dir, samples)
 
     manifest = {
-        "schema_version": 1,
+        "schema_version": 2,
         "description": "GeoIntel operator real-data samples for configured-YOLO QA validation.",
         "output_dir": str(output_dir),
         "sample_width": args.width,
@@ -610,6 +648,8 @@ def main() -> int:
         "half_size_scale": args.half_size_scale,
         "reference_page_limit": args.reference_page_limit,
         "reference_max_features": args.reference_max_features,
+        "default_validation_sample_slugs": sorted(DEFAULT_VALIDATION_SAMPLE_SLUGS),
+        "training_expansion_sample_slugs": sorted(TRAINING_EXPANSION_SAMPLE_SLUGS),
         "samples": samples,
     }
     manifest_path = output_dir / args.manifest_name