Update utils/polygon_utils.py

utils/polygon_utils.py

@@ -34,13 +34,33 @@ def calculate_iou(poly1, poly2):
         print(f"⚠️ calculate_iou failed: {e}")
         return 0.0
 
-
-def calculate_polygon_overlap(ocr_poly, bubble_poly):
+def sanitize_polygon(poly):
     """
-    Calculate what percentage of OCR polygon is inside bubble polygon.
-    overlap = area(ocr ∩ bubble) / area(ocr)
+    Ensures polygon has at least 4 distinct points.
+    Returns None if invalid.
     """
+    if not poly:
+        return None
+
+    # Flatten & cast to int
+    pts = [(int(x), int(y)) for x, y in poly if isinstance(x, (int,float)) and isinstance(y, (int,float))]
+
+    # Remove duplicates
+    pts = list(dict.fromkeys(pts))
+
+    # Must have ≥ 4 points
+    if len(pts) < 4:
+        return None
+
+    return pts
+
+def calculate_polygon_overlap(ocr_poly, bubble_poly):
     try:
+        if not ocr_poly or not bubble_poly:
+            return 0.0
+        if len(ocr_poly) < 4 or len(bubble_poly) < 4:
+            return 0.0
+
         ocr_shape = Polygon(ocr_poly)
         bubble_shape = Polygon(bubble_poly)
 
@@ -49,10 +69,11 @@ def calculate_polygon_overlap(ocr_poly, bubble_poly):
         if not bubble_shape.is_valid:
             bubble_shape = bubble_shape.buffer(0)
 
-        intersection = ocr_shape.intersection(bubble_shape).area
+        inter = ocr_shape.intersection(bubble_shape).area
         ocr_area = ocr_shape.area
 
-        return intersection / ocr_area if ocr_area > 0 else 0.0
+        return inter / ocr_area if ocr_area > 0 else 0.0
+
     except Exception as e:
         print(f"⚠️ calculate_polygon_overlap failed: {e}")
         return 0.0