mineru 2.5.3__py3-none-any.whl → 2.6.0__py3-none-any.whl

mineru/model/utils/pytorchocr/utils/resources/dict/ppocrv5_te_dict.txt

@@ -0,0 +1,540 @@
+!
+"
+#
+$
+%
+&
+'
+(
+)
+*
++
+,
+-
+.
+/
+0
+1
+2
+3
+4
+5
+6
+7
+8
+9
+:
+;
+<
+=
+>
+?
+A
+B
+C
+D
+E
+F
+G
+H
+I
+J
+K
+L
+M
+N
+O
+P
+Q
+R
+S
+T
+U
+V
+W
+X
+Y
+Z
+[
+]
+_
+`
+a
+b
+c
+d
+e
+f
+g
+h
+i
+j
+k
+l
+m
+n
+o
+p
+q
+r
+s
+t
+u
+v
+w
+x
+y
+z
+©
+‥
+{
+}
+\
+|
+@
+^
+~
+÷
+∕
+∙
+⋅
+·
+±
+∓
+∩
+∪
+□
+←
+↔
+⇒
+⇐
+⇔
+∀
+∃
+∄
+∴
+∵
+∝
+∞
+⊥
+∟
+∠
+∡
+∢
+′
+″
+∥
+⊾
+⊿
+∂
+∫
+∬
+∭
+∮
+∯
+∰
+∑
+∏
+√
+∛
+∜
+∱
+∲
+∳
+∶
+∷
+∼
+®
+℉
+Ω
+℧
+Å
+⌀
+ℏ
+⅀
+⍺
+⍵
+¢
+€
+£
+¥
+₿
+Ⅰ
+Ⅱ
+Ⅲ
+Ⅳ
+Ⅴ
+Ⅵ
+Ⅶ
+Ⅷ
+Ⅸ
+Ⅹ
+Ⅺ
+Ⅻ
+ⅰ
+ⅱ
+ⅲ
+ⅳ
+ⅴ
+ⅵ
+ⅶ
+ⅷ
+ⅸ
+ⅹ
+ⅺ
+ⅻ
+➀
+➁
+➂
+➃
+➄
+➅
+➆
+➇
+➈
+➉
+➊
+➋
+➌
+➍
+➎
+➏
+➐
+➑
+➒
+➓
+❶
+❷
+❸
+❹
+❺
+❻
+❼
+❽
+❾
+❿
+①
+②
+③
+④
+⑤
+⑥
+⑦
+⑧
+⑨
+⑩
+●
+▶
+𝑢
+︽
+–
+﹥
+𝜓
+•
+∋
+ƒ
+⬆
+Ạ
+◀
+
+▫
+︾
+À
+Á
+Â
+Ã
+Ä
+Å
+Æ
+Ç
+È
+É
+Ê
+Ë
+Ì
+Í
+Î
+Ï
+Ð
+Ñ
+Ò
+Ó
+Ô
+Õ
+Ö
+Ø
+Ù
+Ú
+Û
+Ü
+Ý
+Þ
+à
+á
+â
+ã
+ä
+å
+æ
+ç
+è
+é
+ê
+ë
+ì
+í
+î
+ï
+ð
+ñ
+ò
+ó
+ô
+õ
+ö
+ø
+ù
+ú
+û
+ü
+ý
+þ
+ÿ
+¡
+¤
+¦
+§
+¨
+ª
+«
+¬
+¯
+°
+²
+³
+´
+µ
+¶
+¸
+¹
+º
+»
+¼
+½
+¾
+¿
+×
+‐
+‑
+‒
+—
+―
+‖
+‗
+‘
+’
+‚
+‛
+“
+”
+„
+‟
+†
+‡
+‣
+․
+…
+‧
+‰
+‴
+‵
+‶
+‷
+‸
+‹
+›
+※
+‼
+‽
+‾
+₤
+₡
+₹
+−
+∖
+∗
+≈
+≠
+≡
+≤
+≥
+⊂
+⊃
+↑
+→
+↓
+↕
+™
+Ω
+℮
+∆
+✓
+✗
+✘
+▪
+◼
+✔
+✕
+☑
+☒
+№
+₽
+₴
+Α
+α
+Β
+β
+Γ
+γ
+Δ
+δ
+Ε
+ε
+Ζ
+ζ
+Η
+η
+Θ
+θ
+Ι
+ι
+Κ
+κ
+Λ
+λ
+Μ
+μ
+Ν
+ν
+Ξ
+ξ
+Ο
+ο
+Π
+π
+Ρ
+ρ
+Σ
+σ
+ς
+Τ
+τ
+Υ
+υ
+Φ
+φ
+Χ
+χ
+Ψ
+ψ
+ω
+ఀ
+ఁ
+ం
+ః
+ఄ
+అ
+ఆ
+ఇ
+ఈ
+ఉ
+ఊ
+ఋ
+ఌ
+ఎ
+ఏ
+ఐ
+ఒ
+ఓ
+ఔ
+క
+ఖ
+గ
+ఘ
+ఙ
+చ
+ఛ
+జ
+ఝ
+ఞ
+ట
+ఠ
+డ
+ఢ
+ణ
+త
+థ
+ద
+ధ
+న
+ప
+ఫ
+బ
+భ
+మ
+య
+ర
+ఱ
+ల
+ళ
+ఴ
+వ
+శ
+ష
+స
+హ
+఼
+ఽ
+ా
+ి
+ీ
+ు
+ూ
+ృ
+ౄ
+ె
+ే
+ై
+ొ
+ో
+ౌ
+్
+ౕ
+ౖ
+ౘ
+ౙ
+ౚ
+ౝ
+ౠ
+ౡ
+ౢ
+ౣ
+౦
+౧
+౨
+౩
+౪
+౫
+౬
+౭
+౮
+౯
+౷
+౸
+౹
+౺
+౻
+౼
+౽
+౾
+౿

mineru/model/{ocr/paddleocr2pytorch → utils}/pytorchocr/utils/resources/models_config.yml

@@ -36,13 +36,13 @@ lang:
     rec: ch_PP-OCRv5_rec_server_infer.pth
     dict: ppocrv5_dict.txt
   ta:
-    det: Multilingual_PP-OCRv3_det_infer.pth
-    rec: ta_PP-OCRv3_rec_infer.pth
-    dict: ta_dict.txt
+    det: ch_PP-OCRv5_det_infer.pth
+    rec: ta_PP-OCRv5_rec_infer.pth
+    dict: ppocrv5_ta_dict.txt
   te:
-    det: Multilingual_PP-OCRv3_det_infer.pth
-    rec: te_PP-OCRv3_rec_infer.pth
-    dict: te_dict.txt
+    det: ch_PP-OCRv5_det_infer.pth
+    rec: te_PP-OCRv5_rec_infer.pth
+    dict: ppocrv5_te_dict.txt
   ka:
     det: Multilingual_PP-OCRv3_det_infer.pth
     rec: ka_PP-OCRv3_rec_infer.pth
@@ -52,17 +52,17 @@ lang:
     rec: latin_PP-OCRv5_rec_infer.pth
     dict: ppocrv5_latin_dict.txt
   arabic:
-    det: Multilingual_PP-OCRv3_det_infer.pth
-    rec: arabic_PP-OCRv3_rec_infer.pth
-    dict: arabic_dict.txt
+    det: ch_PP-OCRv5_det_infer.pth
+    rec: arabic_PP-OCRv5_rec_infer.pth
+    dict: ppocrv5_arabic_dict.txt
   cyrillic:
-    det: Multilingual_PP-OCRv3_det_infer.pth
-    rec: cyrillic_PP-OCRv3_rec_infer.pth
-    dict: cyrillic_dict.txt
+    det: ch_PP-OCRv5_det_infer.pth
+    rec: cyrillic_PP-OCRv5_rec_infer.pth
+    dict: ppocrv5_cyrillic_dict.txt
   devanagari:
-    det: Multilingual_PP-OCRv3_det_infer.pth
-    rec: devanagari_PP-OCRv3_rec_infer.pth
-    dict: devanagari_dict.txt
+    det: ch_PP-OCRv5_det_infer.pth
+    rec: devanagari_PP-OCRv5_rec_infer.pth
+    dict: ppocrv5_devanagari_dict.txt
   east_slavic:
     det: ch_PP-OCRv5_det_infer.pth
     rec: eslav_PP-OCRv5_rec_infer.pth

mineru/model/utils/pytorchocr/utils/resources/pp_formulanet_arch_config.yaml

@@ -0,0 +1,24 @@
+Architecture:
+  model_type: rec
+  algorithm: PP-FormulaNet_plus-M
+  in_channels: 3
+  Transform:
+  Backbone:
+    name: PPHGNetV2_B6_Formula
+    class_num: 1024
+
+  Head:
+    name: PPFormulaNet_Head
+    max_new_tokens: 2560
+    decoder_start_token_id: 0
+    decoder_ffn_dim: 2048
+    decoder_hidden_size: 512
+    decoder_layers: 6
+    temperature: 0.2
+    do_sample: False
+    top_p: 0.95 
+    encoder_hidden_size: 2048
+    is_export: False
+    length_aware: False 
+    use_parallel: False
+    parallel_step: 0

mineru/model/utils/tools/infer/__init__.py

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

mineru/model/{ocr/paddleocr2pytorch → utils}/tools/infer/predict_det.py

@@ -116,6 +116,9 @@ class TextDetector(BaseOCRV20):
         self.load_pytorch_weights(self.weights_path)
         self.net.eval()
         self.net.to(self.device)
+        for module in self.net.modules():
+            if hasattr(module, 'rep'):
+                module.rep()
 
     def _batch_process_same_size(self, img_list):
         """
@@ -293,7 +296,7 @@ class TextDetector(BaseOCRV20):
         return dt_boxes
 
     def __call__(self, img):
-        ori_im = img.copy()
+        ori_shape = img.shape
         data = {'image': img}
         data = transform(data, self.preprocess_op)
         img, shape_list = data
@@ -331,9 +334,9 @@ class TextDetector(BaseOCRV20):
         if (self.det_algorithm == "SAST" and
             self.det_sast_polygon) or (self.det_algorithm in ["PSE", "FCE"] and
                                        self.postprocess_op.box_type == 'poly'):
-            dt_boxes = self.filter_tag_det_res_only_clip(dt_boxes, ori_im.shape)
+            dt_boxes = self.filter_tag_det_res_only_clip(dt_boxes, ori_shape)
         else:
-            dt_boxes = self.filter_tag_det_res(dt_boxes, ori_im.shape)
+            dt_boxes = self.filter_tag_det_res(dt_boxes, ori_shape)
 
         elapse = time.time() - starttime
         return dt_boxes, elapse

mineru/model/{ocr/paddleocr2pytorch → utils}/tools/infer/predict_rec.py

@@ -9,6 +9,7 @@ from tqdm import tqdm
 from ...pytorchocr.base_ocr_v20 import BaseOCRV20
 from . import pytorchocr_utility as utility
 from ...pytorchocr.postprocess import build_post_process
+from ...pytorchocr.modeling.backbones.rec_hgnet import ConvBNAct
 
 
 class TextRecognizer(BaseOCRV20):
@@ -93,6 +94,12 @@ class TextRecognizer(BaseOCRV20):
         self.load_state_dict(weights)
         self.net.eval()
         self.net.to(self.device)
+        for module in self.net.modules():
+            if isinstance(module, ConvBNAct):
+                if module.use_act:
+                    torch.quantization.fuse_modules(module, ['conv', 'bn', 'act'], inplace=True)
+                else:
+                    torch.quantization.fuse_modules(module, ['conv', 'bn'], inplace=True)
 
     def resize_norm_img(self, img, max_wh_ratio):
         imgC, imgH, imgW = self.rec_image_shape
@@ -125,23 +132,15 @@ class TextRecognizer(BaseOCRV20):
 
         assert imgC == img.shape[2]
         max_wh_ratio = max(max_wh_ratio, imgW / imgH)
-        imgW = int((imgH * max_wh_ratio))
+        imgW = int(imgH * max_wh_ratio)
         imgW = max(min(imgW, self.limited_max_width), self.limited_min_width)
         h, w = img.shape[:2]
         ratio = w / float(h)
-        ratio_imgH = math.ceil(imgH * ratio)
-        ratio_imgH = max(ratio_imgH, self.limited_min_width)
-        if ratio_imgH > imgW:
-            resized_w = imgW
-        else:
-            resized_w = int(ratio_imgH)
-        resized_image = cv2.resize(img, (resized_w, imgH))
-        resized_image = resized_image.astype('float32')
-        resized_image = resized_image.transpose((2, 0, 1)) / 255
-        resized_image -= 0.5
-        resized_image /= 0.5
+        ratio_imgH = max(math.ceil(imgH * ratio), self.limited_min_width)
+        resized_w = min(imgW, int(ratio_imgH))
+        resized_image = cv2.resize(img, (resized_w, imgH)) /127.5 - 1
         padding_im = np.zeros((imgC, imgH, imgW), dtype=np.float32)
-        padding_im[:, :, 0:resized_w] = resized_image
+        padding_im[:, :, 0:resized_w] = resized_image.transpose((2, 0, 1))
         return padding_im
 
     def resize_norm_img_svtr(self, img, image_shape):
@@ -307,12 +306,7 @@ class TextRecognizer(BaseOCRV20):
             for beg_img_no in range(0, img_num, batch_num):
                 end_img_no = min(img_num, beg_img_no + batch_num)
                 norm_img_batch = []
-                max_wh_ratio = 0
-                for ino in range(beg_img_no, end_img_no):
-                    # h, w = img_list[ino].shape[0:2]
-                    h, w = img_list[indices[ino]].shape[0:2]
-                    wh_ratio = w * 1.0 / h
-                    max_wh_ratio = max(max_wh_ratio, wh_ratio)
+                max_wh_ratio = width_list[indices[end_img_no - 1]]
                 for ino in range(beg_img_no, end_img_no):
                     if self.rec_algorithm == "SAR":
                         norm_img, _, _, valid_ratio = self.resize_norm_img_sar(
@@ -420,14 +414,11 @@ class TextRecognizer(BaseOCRV20):
                     with torch.no_grad():
                         inp = torch.from_numpy(norm_img_batch)
                         inp = inp.to(self.device)
-                        prob_out = self.net(inp)
+                        preds = self.net(inp)
 
-                    if isinstance(prob_out, list):
-                        preds = [v.cpu().numpy() for v in prob_out]
-                    else:
-                        preds = prob_out.cpu().numpy()
+                with torch.no_grad():
+                    rec_result = self.postprocess_op(preds)
 
-                rec_result = self.postprocess_op(preds)
                 for rno in range(len(rec_result)):
                     rec_res[indices[beg_img_no + rno]] = rec_result[rno]
                 elapse += time.time() - starttime