Update mdr_pdf_parser.py

mdr_pdf_parser.py

@@ -1624,7 +1624,18 @@ class MDROcrEngine:
   def _get_system(self) -> _MDR_TextSystem | None:
     if self._text_system is None:
         paths = {k: str(Path(self._model_dir)/Path(*p)) for k,p in _MDR_OCR_MODELS.items()}
-        self._onnx_params = _MDR_ONNXParams(use_gpu=(self._device=="cpu"), det_model_dir=paths["det"], cls_model_dir=paths["cls"], rec_model_dir=paths["rec"], rec_char_dict_path=paths["keys"])
+        # In MDROcrEngine._get_system()
+        self._onnx_params = _MDR_ONNXParams(
+            use_gpu=(self._device=="cpu"),
+            det_model_dir=paths["det"],
+            cls_model_dir=paths["cls"],
+            rec_model_dir=paths["rec"],
+            rec_char_dict_path=paths["keys"],
+            # --- EXPERIMENT: INCREASE THRESHOLDS ---
+            det_db_thresh=0.5,        # Original was 0.3
+            det_db_box_thresh=0.8,    # Original was 0.6
+            # --- END EXPERIMENT ---
+        )
         try: self._text_system = _MDR_TextSystem(self._onnx_params); print(f"MDR OCR System initialized.")
         except Exception as e: print(f"ERROR initializing MDR OCR System: {e}"); self._text_system = None
     return self._text_system