Add dated regional imagery training sources

2026-07-27 02:39:47 +02:00
parent e607cbe724
commit 65d263a37b
5 changed files with 60 additions and 10 deletions
@@ -88,7 +88,13 @@ def main() -> int:
    manifest = json.loads(args.corpus_manifest.read_text(encoding="utf-8"))
    regions = {item["sample_slug"]: item["region"] for item in manifest["samples"]}
    pure_empty_slugs = {
-        item["sample_slug"] for item in manifest["samples"] if bool(item.get("require_empty"))
+        item["sample_slug"]
+        for item in manifest["samples"]
+        if bool(item.get("require_empty"))
+        or (
+            item.get("sample_role") == "background_candidate"
+            and int(item.get("reference_feature_count") or 0) == 0
+        )
    }
    tiles = [item for item in summary["tiles"] if item.get("kept", True) and item["split"] == args.split]
    image_paths = [item["image_path"] for item in tiles]