feat: add glm ocr

app/services/layout_detector.py

@@ -87,11 +87,11 @@ class LayoutDetector:
     def _get_layout_detector(self):
         """Get or create LayoutDetection instance."""
         if LayoutDetector._layout_detector is None:
-            LayoutDetector._layout_detector = LayoutDetection(model_name="PP-DocLayoutV2")
+            LayoutDetector._layout_detector = LayoutDetection(model_name="PP-DocLayoutV3")
         return LayoutDetector._layout_detector
 
     def detect(self, image: np.ndarray) -> LayoutInfo:
-        """Detect layout of the image using PP-DocLayoutV2.
+        """Detect layout of the image using PP-DocLayoutV3.
 
         Args:
             image: Input image as numpy array.
@@ -125,13 +125,14 @@ class LayoutDetector:
             # Normalize label to region type
             region_type = self.LABEL_TO_TYPE.get(label, "text")
 
-            regions.append(LayoutRegion(
-                type=region_type,
-                bbox=coordinate,
-                confidence=score,
-                score=score,
-            ))
-
+            regions.append(
+                LayoutRegion(
+                    type=region_type,
+                    bbox=coordinate,
+                    confidence=score,
+                    score=score,
+                )
+            )
 
         mixed_recognition = any(region.type == "text" and region.score > 0.85 for region in regions)
 
@@ -144,14 +145,14 @@ if __name__ == "__main__":
     from app.services.image_processor import ImageProcessor
     from app.services.converter import Converter
     from app.services.ocr_service import OCRService
-    
+
     settings = get_settings()
-    
+
     # Initialize dependencies
     layout_detector = LayoutDetector()
     image_processor = ImageProcessor(padding_ratio=settings.image_padding_ratio)
     converter = Converter()
-    
+
     # Initialize OCR service
     ocr_service = OCRService(
         vl_server_url=settings.paddleocr_vl_url,
@@ -159,20 +160,20 @@ if __name__ == "__main__":
         image_processor=image_processor,
         converter=converter,
     )
-    
+
     # Load test image
-    image_path = "test/complex_formula.png"
+    image_path = "test/timeout.jpg"
     image = cv2.imread(image_path)
-    
+
     if image is None:
         print(f"Failed to load image: {image_path}")
     else:
         print(f"Image loaded: {image.shape}")
-        
+
         # Run OCR recognition
         result = ocr_service.recognize(image)
-        
+
         print("\n=== OCR Result ===")
         print(f"Markdown:\n{result['markdown']}")
         print(f"\nLaTeX:\n{result['latex']}")
-        print(f"\nMathML:\n{result['mathml']}")
+        print(f"\nMathML:\n{result['mathml']}")