chore: optimize code

python/rapidocr_onnxruntime/ch_ppocr_det/text_detect.py

@@ -63,15 +63,15 @@ def __call__(self, img: np.ndarray) -> Tuple[Optional[np.ndarray], float]:
         return dt_boxes, elapse
 
     def get_preprocess(self, max_wh):
-        if self.limit_type == 'min':
+        if self.limit_type == "min":
             limit_side_len = self.limit_side_len
         elif max_wh < 960:
             limit_side_len = 960
         elif max_wh < 1500:
             limit_side_len = 1500
         else:
             limit_side_len = 2000
-        return DetPreProcess(limit_side_len, self.limit_type,self.mean, self.std)
+        return DetPreProcess(limit_side_len, self.limit_type, self.mean, self.std)
 
     def filter_tag_det_res(
         self, dt_boxes: np.ndarray, image_shape: Tuple[int, int]

python/rapidocr_onnxruntime/ch_ppocr_det/utils.py

@@ -10,12 +10,15 @@
 
 
 class DetPreProcess:
-    def __init__(self, limit_side_len: int = 736, limit_type: str = "min", mean=None,
-                 std=None):
+    def __init__(
+        self, limit_side_len: int = 736, limit_type: str = "min", mean=None, std=None
+    ):
         if mean is None:
             mean = [0.5, 0.5, 0.5]
+
         if std is None:
             std = [0.5, 0.5, 0.5]
+
         self.mean = np.array(mean)
         self.std = np.array(std)
         self.scale = 1 / 255.0

python/rapidocr_onnxruntime/config.yaml

@@ -26,10 +26,7 @@ Det:
     limit_type: min
     std: [ 0.5, 0.5, 0.5 ]
     mean: [ 0.5, 0.5, 0.5 ]
-#    limit_side_len: 960
-#    limit_type: max
-#    std: [ 0.229, 0.224, 0.225 ]
-#    mean: [ 0.485, 0.456, 0.406 ]
+
     thresh: 0.3
     box_thresh: 0.5
     max_candidates: 1000

python/rapidocr_openvino/ch_ppocr_det/utils.py

@@ -10,12 +10,15 @@
 
 
 class DetPreProcess:
-    def __init__(self, limit_side_len: int = 736, limit_type: str = "min",mean=None,
-                 std=None):
+    def __init__(
+        self, limit_side_len: int = 736, limit_type: str = "min", mean=None, std=None
+    ):
         if mean is None:
-            mean = [0.485, 0.456, 0.406]
+            mean = [0.5, 0.5, 0.5]
+
         if std is None:
-            std = [0.229, 0.224, 0.225]
+            std = [0.5, 0.5, 0.5]
+
         self.mean = np.array(mean)
         self.std = np.array(std)
         self.scale = 1 / 255.0

python/rapidocr_openvino/config.yaml

@@ -21,6 +21,8 @@ Det:
 
     limit_side_len: 736
     limit_type: min
+    std: [ 0.5, 0.5, 0.5 ]
+    mean: [ 0.5, 0.5, 0.5 ]
 
     thresh: 0.3
     box_thresh: 0.5

python/rapidocr_paddle/ch_ppocr_cls/text_cls.py

@@ -11,7 +11,6 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-import argparse
 import copy
 import math
 import time
@@ -20,7 +19,7 @@
 import cv2
 import numpy as np
 
-from rapidocr_paddle.utils import PaddleInferSession, read_yaml
+from rapidocr_paddle.utils import PaddleInferSession
 
 from .utils import ClsPostProcess
 
@@ -96,19 +95,3 @@ def resize_norm_img(self, img):
         padding_im = np.zeros((img_c, img_h, img_w), dtype=np.float32)
         padding_im[:, :, :resized_w] = resized_image
         return padding_im
-
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser()
-    parser.add_argument("--image_path", type=str, help="image_dir|image_path")
-    parser.add_argument("--config_path", type=str, default="config.yaml")
-    args = parser.parse_args()
-
-    config = read_yaml(args.config_path)
-
-    text_classifier = TextClassifier(config)
-
-    img = cv2.imread(args.image_path)
-    img_list, cls_res, predict_time = text_classifier(img)
-    for ino in range(len(img_list)):
-        print(f"cls result:{cls_res[ino]}")

python/rapidocr_paddle/ch_ppocr_det/utils.py

@@ -10,12 +10,15 @@
 
 
 class DetPreProcess:
-    def __init__(self, limit_side_len: int = 736, limit_type: str = "min", mean=None,
-                 std=None):
+    def __init__(
+        self, limit_side_len: int = 736, limit_type: str = "min", mean=None, std=None
+    ):
         if mean is None:
-            mean = [0.485, 0.456, 0.406]
+            mean = [0.5, 0.5, 0.5]
+
         if std is None:
-            std = [0.229, 0.224, 0.225]
+            std = [0.5, 0.5, 0.5]
+
         self.mean = np.array(mean)
         self.std = np.array(std)
         self.scale = 1 / 255.0

python/rapidocr_paddle/ch_ppocr_rec/text_recognize.py

@@ -11,7 +11,6 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-import argparse
 import math
 import time
 from pathlib import Path
@@ -20,7 +19,7 @@
 import cv2
 import numpy as np
 
-from rapidocr_paddle.utils import PaddleInferSession, read_yaml
+from rapidocr_paddle.utils import PaddleInferSession
 
 from .utils import CTCLabelDecode
 
@@ -112,17 +111,3 @@ def resize_norm_img(self, img: np.ndarray, max_wh_ratio: float) -> np.ndarray:
         padding_im = np.zeros((img_channel, img_height, img_width), dtype=np.float32)
         padding_im[:, :, 0:resized_w] = resized_image
         return padding_im
-
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser()
-    parser.add_argument("--image_path", type=str, help="image_dir|image_path")
-    parser.add_argument("--config_path", type=str, default="config.yaml")
-    args = parser.parse_args()
-
-    config = read_yaml(args.config_path)
-    text_recognizer = TextRecognizer(config)
-
-    img = cv2.imread(args.image_path)
-    rec_res, predict_time = text_recognizer(img)
-    print(f"rec result: {rec_res}\t cost: {predict_time}s")

python/rapidocr_paddle/config.yaml

@@ -23,6 +23,8 @@ Det:
 
     limit_side_len: 736
     limit_type: min
+    std: [ 0.5, 0.5, 0.5 ]
+    mean: [ 0.5, 0.5, 0.5 ]
 
     thresh: 0.3
     box_thresh: 0.5

python/tests/test_paddle.py

@@ -1,6 +1,13 @@
 # -*- encoding: utf-8 -*-
 # @Author: SWHL
 # @Contact: liekkaskono@163.com
+"""
+以下测试用例在以下环境中可以全部通过
+OS: macOS 15.1.1 (24B91)
+Python: 3.10.13
+paddlepaddle: 3.0.0b2 CPU
+"""
+
 import sys
 from pathlib import Path
 from typing import List
@@ -27,7 +34,7 @@ def test_long_img():
     download_file(img_url, save_path=img_path)
     result, _ = engine(img_path)
     assert result is not None
-    assert len(result) == 55
+    assert len(result) == 53
     img_path.unlink()
 
 
@@ -62,7 +69,7 @@ def test_transparent_img(img_name: str, gt: str):
         (
             "test_letterbox_like.jpg",
             2,
-            "A：：取决于所使用的执行提供者，它可能没有完全支持模型中的所有操作。回落到CPU操作可能会导致性能速度的下降。此外，即使一个操作是由CUDAeXecution",
+            "A：：取决于所使用的执行提供者，它可能没有完全支持模型中的所有操作。回落到CPU操作可能会导致性能速度的下降。此外，即使一个操作是由CUDAexecution",
         ),
         ("test_without_det.jpg", 1, "在中国作家协会第三届儿童文学"),
     ],
@@ -79,7 +86,7 @@ def test_only_det():
     result, _ = engine(img_path, use_det=True, use_cls=False, use_rec=False)
 
     assert len(result) == 18
-    assert result[0][0] == [5.0, 2.0]
+    assert result[0][0] == [6.0, 2.0]
 
 
 def test_only_cls():
@@ -212,7 +219,7 @@ def test_input_three_ndim_one_channel():
         ),
         (
             "text_vertical_words.png",
-            ["已", "取", "之", "時", "不", "參", "一", "人", "见", "而"],
+            ["已", "取", "之", "時", "不", "參", "一", "人", "見", "而"],
         ),
         (
             "issue_170.png",

python/tests/test_vino.py

@@ -1,6 +1,13 @@
 # -*- encoding: utf-8 -*-
 # @Author: SWHL
 # @Contact: liekkaskono@163.com
+"""
+以下测试用例在以下环境中可以全部通过
+OS: macOS 15.1.1 (24B91)
+Python: 3.10.13
+openvino: 2024.0.0
+"""
+
 import sys
 from pathlib import Path
 from typing import List
@@ -27,7 +34,7 @@ def test_long_img():
     download_file(img_url, save_path=img_path)
     result, _ = engine(img_path)
     assert result is not None
-    assert len(result) == 55
+    assert len(result) == 53
     img_path.unlink()
 
 
@@ -62,7 +69,7 @@ def test_transparent_img(img_name: str, gt: str):
         (
             "test_letterbox_like.jpg",
             2,
-            "A：：取决于所使用的执行提供者，它可能没有完全支持模型中的所有操作。回落到CPU操作可能会导致性能速度的下降。此外，即使一个操作是由CUDAeXecution",
+            "A：：取决于所使用的执行提供者，它可能没有完全支持模型中的所有操作。回落到CPU操作可能会导致性能速度的下降。此外，即使一个操作是由CUDAexecution",
         ),
         ("test_without_det.jpg", 1, "在中国作家协会第三届儿童文学"),
     ],
@@ -79,7 +86,7 @@ def test_only_det():
     result, _ = engine(img_path, use_det=True, use_cls=False, use_rec=False)
 
     assert len(result) == 18
-    assert result[0][0] == [5.0, 2.0]
+    assert result[0][0] == [6.0, 2.0]
 
 
 def test_only_cls():
@@ -212,7 +219,7 @@ def test_input_three_ndim_one_channel():
         ),
         (
             "text_vertical_words.png",
-            ["已", "取", "之", "時", "不", "參", "一", "人", "见", "而"],
+            ["已", "取", "之", "時", "不", "參", "一", "人", "見", "而"],
         ),
         (
             "issue_170.png",