Gate YOLO tile labels by visible ratio

scripts/README.md

@@ -321,6 +321,7 @@ docker exec -it geointel python3 /app/scripts/export_operator_yolo_tile_dataset.
   --tile-size 160 \
   --stride 80 \
   --negative-keep-ratio 1.0 \
+  --min-label-visible-ratio 0.25 \
   --val-samples turnhout,retie,kasterlee_bos \
   --force
 ```
@@ -329,6 +330,9 @@ The tile exporter clips GRB building bounding boxes into each tile, writes
 YOLO labels beside each tile image, keeps a deterministic ratio of empty
 negative tiles, and records `yolo_tile_dataset_summary.json` with
 `positive_tile_count`, `negative_tile_count` and skipped negative tile counts.
+`--min-label-visible-ratio` drops labels where only a small clipped fragment of
+the original building bbox is visible inside the tile; this reduces noisy
+tile-edge labels in overlapping-tile datasets. Use `0` for legacy behavior.
 It remains operator tooling only: no provider fetch, no API mutation and no
 automatic model training.
 
@@ -344,7 +348,8 @@ The audit reads the tile summary and YOLO label files, then writes
 `operator_yolo_dataset_quality_audit.json` and
 `operator_yolo_dataset_quality_audit.md`. It reports positive/background sample
 coverage, train/validation split coverage, repeated hard-negative pressure,
-missing or invalid label rows and normalized box-area signals. Treat
+minimum visible label ratio, missing or invalid label rows and normalized
+box-area signals. Treat
 `needs_attention` as a dataset-design warning, not as a runtime failure: the
 next action is usually more positive AOIs, better validation coverage or more
 unique hard negatives rather than simply extending epochs.