Update utils/bubble_utils.py

utils/bubble_utils.py

@@ -15,6 +15,19 @@ from utils.polygon_utils import (
 )
 from utils.bubble_detect import detect_speech_bubbles_robust
 from utils.u2net_detector import detect_bubbles_u2net
+from utils.bubble_detect_rtdetr import detect_bubbles_rtdetr
+
+def bbox_to_polygon(bbox):
+    """
+    Convert RT-DETR [x1, y1, x2, y2] box into a 4-point polygon.
+    """
+    x1, y1, x2, y2 = bbox
+    return [
+        (int(x1), int(y1)),
+        (int(x2), int(y1)),
+        (int(x2), int(y2)),
+        (int(x1), int(y2)),
+    ]
 
 
 # ======================== Debug Visualization ============================
@@ -79,11 +92,21 @@ def bubble_pipeline_single(file_obj, num_chunks=1, polygon_strategy="hybrid", de
     print(f"📄 bubble_pipeline_single: filename={filename}, size={full_img.size}")
 
     # -------------------------------------------------------
-    # 2) Robust bubble detection
+    # 2) Bubble detection (RT-DETR comic model)
     # -------------------------------------------------------
-    bubble_polygons = detect_speech_bubbles_robust(full_img, min_area=400)
-    # bubble_polygons = detect_bubbles_u2net(full_img)
-    print(f"🔍 Detected {len(bubble_polygons)} speech bubbles")
+    detections = detect_bubbles_rtdetr(full_img)
+    
+    bubble_boxes = [d["bbox"] for d in detections if d["class"] == 0]  # pure bubble shapes
+    bubble_text_boxes = [d["bbox"] for d in detections if d["class"] == 1]  # text inside bubble
+    free_text_boxes = [d["bbox"] for d in detections if d["class"] == 2]  # text outside bubble
+    
+    # Convert bubble boxes → polygons for your polygon-based pipeline
+    bubble_polygons = [bbox_to_polygon(b) for b in bubble_boxes]
+    
+    print(f"🔍 RT-DETR: {len(bubble_polygons)} bubbles, "
+          f"{len(bubble_text_boxes)} bubble-text regions, "
+          f"{len(free_text_boxes)} free-text regions")
+
 
     if debug:
         visualize_all_debug(full_img, [], bubble_polygons,