Reject same-year post-mosaic building labels

2026-07-30 03:35:15 +02:00
parent f7e12230cf
commit 5ac3238c27
2 changed files with 44 additions and 3 deletions
@@ -159,9 +159,16 @@ def normalize(
    decisions: list[dict[str, Any]] = []
    seen: set[str] = set()
    counts: Counter[str] = Counter()
-    imagery_cutoff = (
-        datetime.fromisoformat(imagery_valid_to.replace("Z", "+00:00")) if imagery_valid_to else None
-    )
+    imagery_dates = [
+        datetime.fromisoformat(value.replace("Z", "+00:00"))
+        for value in (imagery_observed_at, imagery_valid_to)
+        if value
+    ]
+    # Annual mosaics expose a validity interval but not the flight date for
+    # each pixel. Using the interval end admits buildings created later in the
+    # same year that are visibly absent from the mosaic. The earliest governed
+    # date is therefore the only conservative training-label cutoff.
+    imagery_cutoff = min(imagery_dates) if imagery_dates else None
    with rasterio.open(raster_path) as raster:
        if raster.crs is None:
            raise SystemExit("Raster CRS is required")
@@ -260,6 +267,8 @@ def normalize(
        "min_label_px": min_label_px,
        "imagery_observed_at": imagery_observed_at,
        "imagery_valid_to": imagery_valid_to,
+        "imagery_feature_creation_cutoff": imagery_cutoff.isoformat() if imagery_cutoff else None,
+        "imagery_feature_creation_cutoff_policy": "earliest_governed_imagery_date",
        "reference_observed_at": reference_observed_at,
        "temporal_mismatch_days": temporal_mismatch_days,
        "temporal_alignment_status": temporal_alignment_status,