Preserve validation manifests in failure sampling

backend/tests/test_failure_driven_yolo_sampling.py

@@ -11,6 +11,17 @@ MODULE = importlib.util.module_from_spec(SPEC)
 SPEC.loader.exec_module(MODULE)
 
 
+def test_dataset_validation_source_preserves_manifest_path(tmp_path: Path):
+    source = tmp_path / "dataset.yaml"
+    source.write_text(
+        "path: /data/source\ntrain: /data/source/train.txt\n"
+        "val: /data/source/internal-val.txt\nnames:\n  0: building\n",
+        encoding="utf-8",
+    )
+
+    assert MODULE.dataset_validation_source(source) == "/data/source/internal-val.txt"
+
+
 def test_sampling_repeats_only_failed_region_train_tiles() -> None:
     manifest = {
         "samples": [

scripts/build_failure_driven_yolo_sampling.py

@@ -19,6 +19,15 @@ def file_sha256(path: Path) -> str:
     return digest.hexdigest()
 
 
+def dataset_validation_source(source_yaml: Path) -> str:
+    """Preserve the source dataset's validation contract verbatim."""
+    for raw_line in source_yaml.read_text(encoding="utf-8").splitlines():
+        key, separator, value = raw_line.partition(":")
+        if separator and key.strip() == "val" and value.strip():
+            return value.strip()
+    raise ValueError(f"Source dataset YAML has no validation source: {source_yaml}")
+
+
 def build_sampling(
     *,
     summary: dict[str, Any],
@@ -126,12 +135,12 @@ def main() -> int:
     train_list = args.output_dir / "train-failure-driven.txt"
     train_list.write_text("\n".join(paths) + "\n", encoding="utf-8")
     source_yaml = args.summary.parent / "dataset.yaml"
-    val_dir = args.summary.parent / "images" / "val"
+    val_source = dataset_validation_source(source_yaml)
     dataset_yaml = args.output_dir / "dataset.yaml"
     dataset_yaml.write_text(
         f"path: {args.output_dir}\n"
         f"train: {train_list}\n"
-        f"val: {val_dir}\n"
+        f"val: {val_source}\n"
         "names:\n  0: building\n",
         encoding="utf-8",
     )