Update mdr_pdf_parser.py

mdr_pdf_parser.py

@@ -1780,10 +1780,12 @@ def mdr_correct_layout_fragments(ocr_engine: 'MDROcrEngine', source_img: Image,
 
 # --- MDR OCR Engine ---
 
-_MDR_OCR_MODELS = {"det": ("ppocrv4", "det", "det.onnx"), "cls": ("ppocrv4", "cls", "cls.onnx"),
-                   "rec": ("ppocrv4", "rec", "rec.onnx"), "keys": ("ch_ppocr_server_v2.0", "ppocr_keys_v1.txt")}
+_MDR_OCR_MODELS = {"det": ("ppocr_onnx", "model", "det_model", "en_PP-OCRv3_det_infer.onnx"),
+                   "cls": ("ppocr_onnx", "model", "cls_model", "ch_ppocr_mobile_v2.0_cls_infer.onnx"),
+                   "rec": ("ppocr_onnx", "model", "rec_model", "en_PP-OCRv3_rec_infer.onnx"),
+                   "keys": ("ppocr_onnx", "ppocr", "utils", "dict", "en_dict.txt")}
 
-_MDR_OCR_URL_BASE = "https://huggingface.co/moskize/OnnxOCR/resolve/main/"
+_MDR_OCR_URL_BASE = "https://raw.githubusercontent.com/Kazuhito00/PaddleOCR-ONNX-Sample/main/"
 
 
 @dataclass
@@ -1797,7 +1799,7 @@ class _MDR_ONNXParams:
 
     # Attributes with default values (Group 1)
     use_angle_cls: bool = True
-    rec_image_shape: str = "3,48,320"
+    rec_image_shape: str = "3,48,640"
     cls_image_shape: str = "3,48,192"
     cls_batch_num: int = 6
     cls_thresh: float = 0.9
@@ -1853,7 +1855,7 @@ class MDROcrEngine:
                 det_model_dir=paths["det"],
                 cls_model_dir=paths["cls"],
                 rec_model_dir=paths["rec"],
-                rec_char_dict_path=paths["keys"],
+                rec_char_dict_path=None,
                 # much lower thresholds so we actually get some candidate masks:
                 det_db_thresh=0.1,
                 det_db_box_thresh=0.3,