audit YOLO geometry and overlapping validation rows

scripts/render_operator_yolo_label_qa_contact_sheets.py

@@ -287,6 +287,9 @@ def draw_tile_card(
         "exact_duplicate": ((0, 220, 255), 5),
         "near_duplicate": ((255, 55, 55), 5),
         "possible_nested": ((255, 0, 220), 5),
+        "extreme_aspect_ratio": ((255, 128, 0), 5),
+        "small_dimension": ((80, 160, 255), 5),
+        "tile_edge": ((57, 255, 20), 5),
     }
     for box_index, box in enumerate(boxes):
         x_center = box["center_x"] * thumb_size
@@ -317,14 +320,17 @@ def draw_tile_card(
     return card
 
 
-def build_relationship_highlights(
+def build_label_highlights(
     summary: dict[str, Any],
 ) -> dict[str, dict[int, str]]:
     """Map audited label rows to their highest-priority visual warning."""
     priorities = {
-        "possible_nested": 1,
-        "near_duplicate": 2,
-        "exact_duplicate": 3,
+        "tile_edge": 1,
+        "small_dimension": 2,
+        "extreme_aspect_ratio": 3,
+        "possible_nested": 4,
+        "near_duplicate": 5,
+        "exact_duplicate": 6,
     }
     highlights: dict[str, dict[int, str]] = {}
     flagged_tiles = summary.get("flagged_tiles") or []
@@ -355,9 +361,30 @@ def build_relationship_highlights(
                 ):
                     tile_highlights[index] = relationship_type
 
+        outliers = tile.get("outliers") or []
+        if not isinstance(outliers, list):
+            continue
+        for outlier in outliers:
+            if not isinstance(outlier, dict):
+                continue
+            category = str(outlier.get("category") or "")
+            index = outlier.get("index")
+            if category not in priorities or not isinstance(index, int) or index < 0:
+                continue
+            existing = tile_highlights.get(index)
+            if existing is None or priorities[category] > priorities[existing]:
+                tile_highlights[index] = category
+
     return highlights
 
 
+def build_relationship_highlights(
+    summary: dict[str, Any],
+) -> dict[str, dict[int, str]]:
+    """Backward-compatible name for callers using relationship manifests."""
+    return build_label_highlights(summary)
+
+
 def image_has_low_visual_variance(image_path: Path, blank_range_threshold: int) -> bool:
     image = Image.open(image_path).convert("L")
     min_value, max_value = image.getextrema()
@@ -406,8 +433,8 @@ def build_report(
     invalid_label_count = 0
     valid_label_count = 0
     low_visual_variance_tile_count = 0
-    relationship_highlight_count = 0
-    relationship_highlights = build_relationship_highlights(summary)
+    highlight_category_counts: dict[str, int] = {}
+    label_highlights = build_label_highlights(summary)
 
     for tile in selected_tiles:
         image_path = resolve_path(tile.get("image_path"), summary_path)
@@ -415,8 +442,11 @@ def build_report(
         boxes, tile_invalid_count, missing_label_file = parse_yolo_label_file(
             label_path
         )
-        tile_relationship_highlights = relationship_highlights.get(str(label_path), {})
-        relationship_highlight_count += len(tile_relationship_highlights)
+        tile_label_highlights = label_highlights.get(str(label_path), {})
+        for category in tile_label_highlights.values():
+            highlight_category_counts[category] = (
+                highlight_category_counts.get(category, 0) + 1
+            )
         invalid_label_count += tile_invalid_count
         valid_label_count += len(boxes)
         if missing_label_file:
@@ -441,7 +471,7 @@ def build_report(
                     invalid_label_count=tile_invalid_count,
                     missing_label_file=missing_label_file,
                     low_visual_variance=low_visual_variance,
-                    relationship_highlights=tile_relationship_highlights,
+                    relationship_highlights=tile_label_highlights,
                 )
             )
             rendered = True
@@ -460,7 +490,7 @@ def build_report(
                 "invalid_label_count": tile_invalid_count,
                 "missing_label_file": missing_label_file,
                 "low_visual_variance": low_visual_variance,
-                "relationship_highlight_count": len(tile_relationship_highlights),
+                "highlighted_label_count": len(tile_label_highlights),
                 "rendered": rendered,
             }
         )
@@ -478,6 +508,14 @@ def build_report(
         }
         for start in range(0, len(rendered_cards), args.tiles_per_sheet)
     ]
+    relationship_highlight_count = sum(
+        highlight_category_counts.get(category, 0)
+        for category in ("exact_duplicate", "near_duplicate", "possible_nested")
+    )
+    outlier_highlight_count = sum(
+        highlight_category_counts.get(category, 0)
+        for category in ("extreme_aspect_ratio", "small_dimension", "tile_edge")
+    )
 
     return (
         {
@@ -503,7 +541,12 @@ def build_report(
             "invalid_label_count": invalid_label_count,
             "valid_label_count": valid_label_count,
             "low_visual_variance_tile_count": low_visual_variance_tile_count,
+            "highlighted_label_count": sum(highlight_category_counts.values()),
             "relationship_highlight_count": relationship_highlight_count,
+            "outlier_highlight_count": outlier_highlight_count,
+            "highlight_category_counts": dict(
+                sorted(highlight_category_counts.items())
+            ),
             "blank_range_threshold": args.blank_range_threshold,
             "contact_sheets": contact_sheets,
             "selected_tiles": selected_report_tiles,
@@ -525,7 +568,8 @@ def write_markdown(report: dict[str, Any], output_dir: Path) -> None:
         f"- invalid label rows: {report['invalid_label_count']}",
         f"- low-variance rendered tiles: {report['low_visual_variance_tile_count']}",
         f"- valid labels rendered: {report['valid_label_count']}",
-        f"- relationship-highlighted labels: {report['relationship_highlight_count']}",
+        f"- highlighted labels: {report['highlighted_label_count']}",
+        f"- highlight categories: `{report['highlight_category_counts']}`",
         "",
         "## Contact Sheets",
         "",