magic-pdf 1.2.1__py3-none-any.whl → 1.3.0__py3-none-any.whl

magic_pdf/model/sub_modules/model_init.py

@@ -5,47 +5,57 @@ from magic_pdf.config.constants import MODEL_NAME
 from magic_pdf.model.model_list import AtomicModel
 from magic_pdf.model.sub_modules.language_detection.yolov11.YOLOv11 import YOLOv11LangDetModel
 from magic_pdf.model.sub_modules.layout.doclayout_yolo.DocLayoutYOLO import DocLayoutYOLOModel
-from magic_pdf.model.sub_modules.layout.layoutlmv3.model_init import Layoutlmv3_Predictor
 from magic_pdf.model.sub_modules.mfd.yolov8.YOLOv8 import YOLOv8MFDModel
 from magic_pdf.model.sub_modules.mfr.unimernet.Unimernet import UnimernetModel
-
-try:
-    from magic_pdf_ascend_plugin.libs.license_verifier import load_license, LicenseFormatError, LicenseSignatureError, LicenseExpiredError
-    from magic_pdf_ascend_plugin.model_plugin.ocr.paddleocr.ppocr_273_npu import ModifiedPaddleOCR
-    from magic_pdf_ascend_plugin.model_plugin.table.rapidtable.rapid_table_npu import RapidTableModel
-    license_key = load_license()
-    logger.info(f'Using Ascend Plugin Success, License id is {license_key["payload"]["id"]},'
-                f' License expired at {license_key["payload"]["date"]["end_date"]}')
-except Exception as e:
-    if isinstance(e, ImportError):
-        pass
-    elif isinstance(e, LicenseFormatError):
-        logger.error("Ascend Plugin: Invalid license format. Please check the license file.")
-    elif isinstance(e, LicenseSignatureError):
-        logger.error("Ascend Plugin: Invalid signature. The license may be tampered with.")
-    elif isinstance(e, LicenseExpiredError):
-        logger.error("Ascend Plugin: License has expired. Please renew your license.")
-    elif isinstance(e, FileNotFoundError):
-        logger.error("Ascend Plugin: Not found License file.")
-    else:
-        logger.error(f"Ascend Plugin: {e}")
-    from magic_pdf.model.sub_modules.ocr.paddleocr.ppocr_273_mod import ModifiedPaddleOCR
-    # from magic_pdf.model.sub_modules.ocr.paddleocr.ppocr_291_mod import ModifiedPaddleOCR
-    from magic_pdf.model.sub_modules.table.rapidtable.rapid_table import RapidTableModel
-
-from magic_pdf.model.sub_modules.table.structeqtable.struct_eqtable import StructTableModel
-from magic_pdf.model.sub_modules.table.tablemaster.tablemaster_paddle import TableMasterPaddleModel
-
-def table_model_init(table_model_type, model_path, max_time, _device_='cpu', ocr_engine=None, table_sub_model_name=None):
+from magic_pdf.model.sub_modules.ocr.paddleocr2pytorch.pytorch_paddle import PytorchPaddleOCR
+from magic_pdf.model.sub_modules.table.rapidtable.rapid_table import RapidTableModel
+# try:
+#     from magic_pdf_ascend_plugin.libs.license_verifier import (
+#         LicenseExpiredError, LicenseFormatError, LicenseSignatureError,
+#         load_license)
+#     from magic_pdf_ascend_plugin.model_plugin.ocr.paddleocr.ppocr_273_npu import ModifiedPaddleOCR
+#     from magic_pdf_ascend_plugin.model_plugin.table.rapidtable.rapid_table_npu import RapidTableModel
+#     license_key = load_license()
+#     logger.info(f'Using Ascend Plugin Success, License id is {license_key["payload"]["id"]},'
+#                 f' License expired at {license_key["payload"]["date"]["end_date"]}')
+# except Exception as e:
+#     if isinstance(e, ImportError):
+#         pass
+#     elif isinstance(e, LicenseFormatError):
+#         logger.error('Ascend Plugin: Invalid license format. Please check the license file.')
+#     elif isinstance(e, LicenseSignatureError):
+#         logger.error('Ascend Plugin: Invalid signature. The license may be tampered with.')
+#     elif isinstance(e, LicenseExpiredError):
+#         logger.error('Ascend Plugin: License has expired. Please renew your license.')
+#     elif isinstance(e, FileNotFoundError):
+#         logger.error('Ascend Plugin: Not found License file.')
+#     else:
+#         logger.error(f'Ascend Plugin: {e}')
+#     from magic_pdf.model.sub_modules.ocr.paddleocr.ppocr_273_mod import ModifiedPaddleOCR
+#     # from magic_pdf.model.sub_modules.ocr.paddleocr.ppocr_291_mod import ModifiedPaddleOCR
+#     from magic_pdf.model.sub_modules.table.rapidtable.rapid_table import RapidTableModel
+
+
+def table_model_init(table_model_type, model_path, max_time, _device_='cpu', lang=None, table_sub_model_name=None):
     if table_model_type == MODEL_NAME.STRUCT_EQTABLE:
+        from magic_pdf.model.sub_modules.table.structeqtable.struct_eqtable import StructTableModel
         table_model = StructTableModel(model_path, max_new_tokens=2048, max_time=max_time)
     elif table_model_type == MODEL_NAME.TABLE_MASTER:
+        from magic_pdf.model.sub_modules.table.tablemaster.tablemaster_paddle import TableMasterPaddleModel
         config = {
             'model_dir': model_path,
             'device': _device_
         }
         table_model = TableMasterPaddleModel(config)
     elif table_model_type == MODEL_NAME.RAPID_TABLE:
+        atom_model_manager = AtomModelSingleton()
+        ocr_engine = atom_model_manager.get_atom_model(
+            atom_model_name='ocr',
+            ocr_show_log=False,
+            det_db_box_thresh=0.5,
+            det_db_unclip_ratio=1.6,
+            lang=lang
+        )
         table_model = RapidTableModel(ocr_engine, table_sub_model_name)
     else:
         logger.error('table model type not allow')
@@ -55,7 +65,7 @@ def table_model_init(table_model_type, model_path, max_time, _device_='cpu', ocr
 
 
 def mfd_model_init(weight, device='cpu'):
-    if str(device).startswith("npu"):
+    if str(device).startswith('npu'):
         device = torch.device(device)
     mfd_model = YOLOv8MFDModel(weight, device)
     return mfd_model
@@ -67,19 +77,20 @@ def mfr_model_init(weight_dir, cfg_path, device='cpu'):
 
 
 def layout_model_init(weight, config_file, device):
+    from magic_pdf.model.sub_modules.layout.layoutlmv3.model_init import Layoutlmv3_Predictor
     model = Layoutlmv3_Predictor(weight, config_file, device)
     return model
 
 
 def doclayout_yolo_model_init(weight, device='cpu'):
-    if str(device).startswith("npu"):
+    if str(device).startswith('npu'):
         device = torch.device(device)
     model = DocLayoutYOLOModel(weight, device)
     return model
 
 
 def langdetect_model_init(langdetect_model_weight, device='cpu'):
-    if str(device).startswith("npu"):
+    if str(device).startswith('npu'):
         device = torch.device(device)
     model = YOLOv11LangDetModel(langdetect_model_weight, device)
     return model
@@ -92,7 +103,8 @@ def ocr_model_init(show_log: bool = False,
                    det_db_unclip_ratio=1.8,
                    ):
     if lang is not None and lang != '':
-        model = ModifiedPaddleOCR(
+        # model = ModifiedPaddleOCR(
+        model = PytorchPaddleOCR(
             show_log=show_log,
             det_db_box_thresh=det_db_box_thresh,
             lang=lang,
@@ -100,7 +112,8 @@ def ocr_model_init(show_log: bool = False,
             det_db_unclip_ratio=det_db_unclip_ratio,
         )
     else:
-        model = ModifiedPaddleOCR(
+        # model = ModifiedPaddleOCR(
+        model = PytorchPaddleOCR(
             show_log=show_log,
             det_db_box_thresh=det_db_box_thresh,
             use_dilation=use_dilation,
@@ -129,7 +142,7 @@ class AtomModelSingleton:
         elif atom_model_name in [AtomicModel.Layout]:
             key = (atom_model_name, layout_model_name)
         elif atom_model_name in [AtomicModel.Table]:
-            key = (atom_model_name, table_model_name)
+            key = (atom_model_name, table_model_name, lang)
         else:
             key = atom_model_name
 
@@ -177,7 +190,7 @@ def atom_model_init(model_name: str, **kwargs):
             kwargs.get('table_model_path'),
             kwargs.get('table_max_time'),
             kwargs.get('device'),
-            kwargs.get('ocr_engine'),
+            kwargs.get('lang'),
             kwargs.get('table_sub_model_name')
         )
     elif model_name == AtomicModel.LangDetect:

magic_pdf/model/sub_modules/model_utils.py

@@ -1,25 +1,31 @@
 import time
-
 import torch
-from PIL import Image
 from loguru import logger
-
+import numpy as np
 from magic_pdf.libs.clean_memory import clean_memory
 
 
-def crop_img(input_res, input_pil_img, crop_paste_x=0, crop_paste_y=0):
+def crop_img(input_res, input_np_img, crop_paste_x=0, crop_paste_y=0):
+
     crop_xmin, crop_ymin = int(input_res['poly'][0]), int(input_res['poly'][1])
     crop_xmax, crop_ymax = int(input_res['poly'][4]), int(input_res['poly'][5])
-    # Create a white background with an additional width and height of 50
+
+    # Calculate new dimensions
     crop_new_width = crop_xmax - crop_xmin + crop_paste_x * 2
     crop_new_height = crop_ymax - crop_ymin + crop_paste_y * 2
-    return_image = Image.new('RGB', (crop_new_width, crop_new_height), 'white')
 
-    # Crop image
-    crop_box = (crop_xmin, crop_ymin, crop_xmax, crop_ymax)
-    cropped_img = input_pil_img.crop(crop_box)
-    return_image.paste(cropped_img, (crop_paste_x, crop_paste_y))
-    return_list = [crop_paste_x, crop_paste_y, crop_xmin, crop_ymin, crop_xmax, crop_ymax, crop_new_width, crop_new_height]
+    # Create a white background array
+    return_image = np.ones((crop_new_height, crop_new_width, 3), dtype=np.uint8) * 255
+
+    # Crop the original image using numpy slicing
+    cropped_img = input_np_img[crop_ymin:crop_ymax, crop_xmin:crop_xmax]
+
+    # Paste the cropped image onto the white background
+    return_image[crop_paste_y:crop_paste_y + (crop_ymax - crop_ymin),
+    crop_paste_x:crop_paste_x + (crop_xmax - crop_xmin)] = cropped_img
+
+    return_list = [crop_paste_x, crop_paste_y, crop_xmin, crop_ymin, crop_xmax, crop_ymax, crop_new_width,
+                   crop_new_height]
     return return_image, return_list
 
 

magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/__init__.py

@@ -0,0 +1 @@
+# Copyright (c) Opendatalab. All rights reserved.

magic_pdf/model/sub_modules/ocr/{paddleocr → paddleocr2pytorch}/ocr_utils.py

@@ -1,58 +1,67 @@
+# Copyright (c) Opendatalab. All rights reserved.
+import copy
+
 import cv2
 import numpy as np
-from loguru import logger
-from io import BytesIO
-from PIL import Image
-import base64
-from magic_pdf.libs.boxbase import __is_overlaps_y_exceeds_threshold
 from magic_pdf.pre_proc.ocr_dict_merge import merge_spans_to_line
-
-import importlib.resources
-from paddleocr import PaddleOCR
-from ppocr.utils.utility import check_and_read
+from magic_pdf.libs.boxbase import __is_overlaps_y_exceeds_threshold
 
 
 def img_decode(content: bytes):
     np_arr = np.frombuffer(content, dtype=np.uint8)
     return cv2.imdecode(np_arr, cv2.IMREAD_UNCHANGED)
 
-
 def check_img(img):
     if isinstance(img, bytes):
         img = img_decode(img)
-    if isinstance(img, str):
-        image_file = img
-        img, flag_gif, flag_pdf = check_and_read(image_file)
-        if not flag_gif and not flag_pdf:
-            with open(image_file, 'rb') as f:
-                img_str = f.read()
-                img = img_decode(img_str)
-            if img is None:
-                try:
-                    buf = BytesIO()
-                    image = BytesIO(img_str)
-                    im = Image.open(image)
-                    rgb = im.convert('RGB')
-                    rgb.save(buf, 'jpeg')
-                    buf.seek(0)
-                    image_bytes = buf.read()
-                    data_base64 = str(base64.b64encode(image_bytes),
-                                      encoding="utf-8")
-                    image_decode = base64.b64decode(data_base64)
-                    img_array = np.frombuffer(image_decode, np.uint8)
-                    img = cv2.imdecode(img_array, cv2.IMREAD_COLOR)
-                except:
-                    logger.error("error in loading image:{}".format(image_file))
-                    return None
-        if img is None:
-            logger.error("error in loading image:{}".format(image_file))
-            return None
     if isinstance(img, np.ndarray) and len(img.shape) == 2:
         img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
+    return img
 
+
+def alpha_to_color(img, alpha_color=(255, 255, 255)):
+    if len(img.shape) == 3 and img.shape[2] == 4:
+        B, G, R, A = cv2.split(img)
+        alpha = A / 255
+
+        R = (alpha_color[0] * (1 - alpha) + R * alpha).astype(np.uint8)
+        G = (alpha_color[1] * (1 - alpha) + G * alpha).astype(np.uint8)
+        B = (alpha_color[2] * (1 - alpha) + B * alpha).astype(np.uint8)
+
+        img = cv2.merge((B, G, R))
     return img
 
 
+def preprocess_image(_image):
+    alpha_color = (255, 255, 255)
+    _image = alpha_to_color(_image, alpha_color)
+    return _image
+
+
+def sorted_boxes(dt_boxes):
+    """
+    Sort text boxes in order from top to bottom, left to right
+    args:
+        dt_boxes(array):detected text boxes with shape [4, 2]
+    return:
+        sorted boxes(array) with shape [4, 2]
+    """
+    num_boxes = dt_boxes.shape[0]
+    sorted_boxes = sorted(dt_boxes, key=lambda x: (x[0][1], x[0][0]))
+    _boxes = list(sorted_boxes)
+
+    for i in range(num_boxes - 1):
+        for j in range(i, -1, -1):
+            if abs(_boxes[j + 1][0][1] - _boxes[j][0][1]) < 10 and \
+                    (_boxes[j + 1][0][0] < _boxes[j][0][0]):
+                tmp = _boxes[j]
+                _boxes[j] = _boxes[j + 1]
+                _boxes[j + 1] = tmp
+            else:
+                break
+    return _boxes
+
+
 def bbox_to_points(bbox):
     """ 将bbox格式转换为四个顶点的数组 """
     x0, y0, x1, y1 = bbox
@@ -252,9 +261,10 @@ def get_adjusted_mfdetrec_res(single_page_mfdetrec_res, useful_list):
     return adjusted_mfdetrec_res
 
 
-def get_ocr_result_list(ocr_res, useful_list):
+def get_ocr_result_list(ocr_res, useful_list, ocr_enable, new_image, lang):
     paste_x, paste_y, xmin, ymin, xmax, ymax, new_width, new_height = useful_list
     ocr_result_list = []
+    ori_im = new_image.copy()
     for box_ocr_res in ocr_res:
 
         if len(box_ocr_res) == 2:
@@ -266,6 +276,11 @@ def get_ocr_result_list(ocr_res, useful_list):
         else:
             p1, p2, p3, p4 = box_ocr_res
             text, score = "", 1
+
+            if ocr_enable:
+                tmp_box = copy.deepcopy(np.array([p1, p2, p3, p4]).astype('float32'))
+                img_crop = get_rotate_crop_image(ori_im, tmp_box)
+
         # average_angle_degrees = calculate_angle_degrees(box_ocr_res[0])
         # if average_angle_degrees > 0.5:
         poly = [p1, p2, p3, p4]
@@ -288,12 +303,22 @@ def get_ocr_result_list(ocr_res, useful_list):
         p3 = [p3[0] - paste_x + xmin, p3[1] - paste_y + ymin]
         p4 = [p4[0] - paste_x + xmin, p4[1] - paste_y + ymin]
 
-        ocr_result_list.append({
-            'category_id': 15,
-            'poly': p1 + p2 + p3 + p4,
-            'score': float(round(score, 2)),
-            'text': text,
-        })
+        if ocr_enable:
+            ocr_result_list.append({
+                'category_id': 15,
+                'poly': p1 + p2 + p3 + p4,
+                'score': 1,
+                'text': text,
+                'np_img': img_crop,
+                'lang': lang,
+            })
+        else:
+            ocr_result_list.append({
+                'category_id': 15,
+                'poly': p1 + p2 + p3 + p4,
+                'score': float(round(score, 2)),
+                'text': text,
+            })
 
     return ocr_result_list
 
@@ -308,56 +333,36 @@ def calculate_is_angle(poly):
         return True
 
 
-class ONNXModelSingleton:
-    _instance = None
-    _models = {}
-
-    def __new__(cls, *args, **kwargs):
-        if cls._instance is None:
-            cls._instance = super().__new__(cls)
-        return cls._instance
-
-    def get_onnx_model(self, **kwargs):
-
-        lang = kwargs.get('lang', None)
-        det_db_box_thresh = kwargs.get('det_db_box_thresh', 0.3)
-        use_dilation = kwargs.get('use_dilation', True)
-        det_db_unclip_ratio = kwargs.get('det_db_unclip_ratio', 1.8)
-        key = (lang, det_db_box_thresh, use_dilation, det_db_unclip_ratio)
-        if key not in self._models:
-            self._models[key] = onnx_model_init(key)
-        return self._models[key]
-
-def onnx_model_init(key):
-    if len(key) < 4:
-        logger.error('Invalid key length, expected at least 4 elements')
-        exit(1)
-
-    try:
-        with importlib.resources.path('rapidocr_onnxruntime.models', '') as resource_path:
-            additional_ocr_params = {
-                "use_onnx": True,
-                "det_model_dir": f'{resource_path}/ch_PP-OCRv4_det_infer.onnx',
-                "rec_model_dir": f'{resource_path}/ch_PP-OCRv4_rec_infer.onnx',
-                "cls_model_dir": f'{resource_path}/ch_ppocr_mobile_v2.0_cls_infer.onnx',
-                "det_db_box_thresh": key[1],
-                "use_dilation": key[2],
-                "det_db_unclip_ratio": key[3],
-            }
-
-            if key[0] is not None:
-                additional_ocr_params["lang"] = key[0]
-
-            # logger.info(f"additional_ocr_params: {additional_ocr_params}")
-
-            onnx_model = PaddleOCR(**additional_ocr_params)
-
-            if onnx_model is None:
-                logger.error('model init failed')
-                exit(1)
-            else:
-                return onnx_model
-
-    except Exception as e:
-        logger.exception(f'Error initializing model: {e}')
-        exit(1)
+def get_rotate_crop_image(img, points):
+    '''
+    img_height, img_width = img.shape[0:2]
+    left = int(np.min(points[:, 0]))
+    right = int(np.max(points[:, 0]))
+    top = int(np.min(points[:, 1]))
+    bottom = int(np.max(points[:, 1]))
+    img_crop = img[top:bottom, left:right, :].copy()
+    points[:, 0] = points[:, 0] - left
+    points[:, 1] = points[:, 1] - top
+    '''
+    assert len(points) == 4, "shape of points must be 4*2"
+    img_crop_width = int(
+        max(
+            np.linalg.norm(points[0] - points[1]),
+            np.linalg.norm(points[2] - points[3])))
+    img_crop_height = int(
+        max(
+            np.linalg.norm(points[0] - points[3]),
+            np.linalg.norm(points[1] - points[2])))
+    pts_std = np.float32([[0, 0], [img_crop_width, 0],
+                          [img_crop_width, img_crop_height],
+                          [0, img_crop_height]])
+    M = cv2.getPerspectiveTransform(points, pts_std)
+    dst_img = cv2.warpPerspective(
+        img,
+        M, (img_crop_width, img_crop_height),
+        borderMode=cv2.BORDER_REPLICATE,
+        flags=cv2.INTER_CUBIC)
+    dst_img_height, dst_img_width = dst_img.shape[0:2]
+    if dst_img_height * 1.0 / dst_img_width >= 1.5:
+        dst_img = np.rot90(dst_img)
+    return dst_img

magic_pdf/model/sub_modules/ocr/paddleocr2pytorch/pytorch_paddle.py

@@ -0,0 +1,193 @@
+# Copyright (c) Opendatalab. All rights reserved.
+import copy
+import os.path
+import warnings
+from pathlib import Path
+
+import cv2
+import numpy as np
+import yaml
+from loguru import logger
+
+from magic_pdf.libs.config_reader import get_device, get_local_models_dir
+from .ocr_utils import check_img, preprocess_image, sorted_boxes, merge_det_boxes, update_det_boxes, get_rotate_crop_image
+from .tools.infer.predict_system import TextSystem
+from .tools.infer import pytorchocr_utility as utility
+import argparse
+
+
+latin_lang = [
+        'af', 'az', 'bs', 'cs', 'cy', 'da', 'de', 'es', 'et', 'fr', 'ga', 'hr',  # noqa: E126
+        'hu', 'id', 'is', 'it', 'ku', 'la', 'lt', 'lv', 'mi', 'ms', 'mt', 'nl',
+        'no', 'oc', 'pi', 'pl', 'pt', 'ro', 'rs_latin', 'sk', 'sl', 'sq', 'sv',
+        'sw', 'tl', 'tr', 'uz', 'vi', 'french', 'german'
+]
+arabic_lang = ['ar', 'fa', 'ug', 'ur']
+cyrillic_lang = [
+        'ru', 'rs_cyrillic', 'be', 'bg', 'uk', 'mn', 'abq', 'ady', 'kbd', 'ava',  # noqa: E126
+        'dar', 'inh', 'che', 'lbe', 'lez', 'tab'
+]
+devanagari_lang = [
+        'hi', 'mr', 'ne', 'bh', 'mai', 'ang', 'bho', 'mah', 'sck', 'new', 'gom',  # noqa: E126
+        'sa', 'bgc'
+]
+
+
+def get_model_params(lang, config):
+    if lang in config['lang']:
+        params = config['lang'][lang]
+        det = params.get('det')
+        rec = params.get('rec')
+        dict_file = params.get('dict')
+        return det, rec, dict_file
+    else:
+        raise Exception (f'Language {lang} not supported')
+
+
+root_dir = Path(__file__).resolve().parent
+
+
+class PytorchPaddleOCR(TextSystem):
+    def __init__(self, *args, **kwargs):
+        parser = utility.init_args()
+        args = parser.parse_args(args)
+
+        self.lang = kwargs.get('lang', 'ch')
+        if self.lang in latin_lang:
+            self.lang = 'latin'
+        elif self.lang in arabic_lang:
+            self.lang = 'arabic'
+        elif self.lang in cyrillic_lang:
+            self.lang = 'cyrillic'
+        elif self.lang in devanagari_lang:
+            self.lang = 'devanagari'
+        else:
+            pass
+
+        models_config_path = os.path.join(root_dir, 'pytorchocr', 'utils', 'resources', 'models_config.yml')
+        with open(models_config_path) as file:
+            config = yaml.safe_load(file)
+            det, rec, dict_file = get_model_params(self.lang, config)
+        ocr_models_dir = os.path.join(get_local_models_dir(), 'OCR', 'paddleocr_torch')
+        kwargs['det_model_path'] = os.path.join(ocr_models_dir, det)
+        kwargs['rec_model_path'] = os.path.join(ocr_models_dir, rec)
+        kwargs['rec_char_dict_path'] = os.path.join(root_dir, 'pytorchocr', 'utils', 'resources', 'dict', dict_file)
+        # kwargs['rec_batch_num'] = 8
+
+        kwargs['device'] = get_device()
+
+        default_args = vars(args)
+        default_args.update(kwargs)
+        args = argparse.Namespace(**default_args)
+
+        super().__init__(args)
+
+    def ocr(self,
+            img,
+            det=True,
+            rec=True,
+            mfd_res=None,
+            tqdm_enable=False,
+            ):
+        assert isinstance(img, (np.ndarray, list, str, bytes))
+        if isinstance(img, list) and det == True:
+            logger.error('When input a list of images, det must be false')
+            exit(0)
+        img = check_img(img)
+        imgs = [img]
+        with warnings.catch_warnings():
+            warnings.simplefilter("ignore", category=RuntimeWarning)
+            if det and rec:
+                ocr_res = []
+                for img in imgs:
+                    img = preprocess_image(img)
+                    dt_boxes, rec_res = self.__call__(img, mfd_res=mfd_res)
+                    if not dt_boxes and not rec_res:
+                        ocr_res.append(None)
+                        continue
+                    tmp_res = [[box.tolist(), res] for box, res in zip(dt_boxes, rec_res)]
+                    ocr_res.append(tmp_res)
+                return ocr_res
+            elif det and not rec:
+                ocr_res = []
+                for img in imgs:
+                    img = preprocess_image(img)
+                    dt_boxes, elapse = self.text_detector(img)
+                    # logger.debug("dt_boxes num : {}, elapsed : {}".format(len(dt_boxes), elapse))
+                    if dt_boxes is None:
+                        ocr_res.append(None)
+                        continue
+                    dt_boxes = sorted_boxes(dt_boxes)
+                    # merge_det_boxes 和 update_det_boxes 都会把poly转成bbox再转回poly，因此需要过滤所有倾斜程度较大的文本框
+                    dt_boxes = merge_det_boxes(dt_boxes)
+                    if mfd_res:
+                        dt_boxes = update_det_boxes(dt_boxes, mfd_res)
+                    tmp_res = [box.tolist() for box in dt_boxes]
+                    ocr_res.append(tmp_res)
+                return ocr_res
+            elif not det and rec:
+                ocr_res = []
+                for img in imgs:
+                    if not isinstance(img, list):
+                        img = preprocess_image(img)
+                        img = [img]
+                    rec_res, elapse = self.text_recognizer(img, tqdm_enable=tqdm_enable)
+                    # logger.debug("rec_res num  : {}, elapsed : {}".format(len(rec_res), elapse))
+                    ocr_res.append(rec_res)
+                return ocr_res
+
+    def __call__(self, img, mfd_res=None):
+
+        if img is None:
+            logger.debug("no valid image provided")
+            return None, None
+
+        ori_im = img.copy()
+        dt_boxes, elapse = self.text_detector(img)
+
+        if dt_boxes is None:
+            logger.debug("no dt_boxes found, elapsed : {}".format(elapse))
+            return None, None
+        else:
+            pass
+            # logger.debug("dt_boxes num : {}, elapsed : {}".format(len(dt_boxes), elapse))
+        img_crop_list = []
+
+        dt_boxes = sorted_boxes(dt_boxes)
+
+        # merge_det_boxes 和 update_det_boxes 都会把poly转成bbox再转回poly，因此需要过滤所有倾斜程度较大的文本框
+        dt_boxes = merge_det_boxes(dt_boxes)
+
+        if mfd_res:
+            dt_boxes = update_det_boxes(dt_boxes, mfd_res)
+
+        for bno in range(len(dt_boxes)):
+            tmp_box = copy.deepcopy(dt_boxes[bno])
+            img_crop = get_rotate_crop_image(ori_im, tmp_box)
+            img_crop_list.append(img_crop)
+
+        rec_res, elapse = self.text_recognizer(img_crop_list)
+        # logger.debug("rec_res num  : {}, elapsed : {}".format(len(rec_res), elapse))
+
+        filter_boxes, filter_rec_res = [], []
+        for box, rec_result in zip(dt_boxes, rec_res):
+            text, score = rec_result
+            if score >= self.drop_score:
+                filter_boxes.append(box)
+                filter_rec_res.append(rec_result)
+
+        return filter_boxes, filter_rec_res
+
+if __name__ == '__main__':
+    pytorch_paddle_ocr = PytorchPaddleOCR()
+    img = cv2.imread("/Users/myhloli/Downloads/screenshot-20250326-194348.png")
+    dt_boxes, rec_res = pytorch_paddle_ocr(img)
+    ocr_res = []
+    if not dt_boxes and not rec_res:
+        ocr_res.append(None)
+    else:
+        tmp_res = [[box.tolist(), res] for box, res in zip(dt_boxes, rec_res)]
+        ocr_res.append(tmp_res)
+    print(ocr_res)
+
+