完成ReCTS转换

WangYX-TKZ · Mar 22, 2020 · 5ba9f29 · 5ba9f29
1 parent 7b8a4e3
commit 5ba9f29
Show file tree

Hide file tree

Showing 8 changed files with 92 additions and 14 deletions.
diff --git a/README.md b/README.md
@@ -5,7 +5,7 @@
     - [x] icdar2015
     - [x] MLT2019
     - [x] COCO-Text_v2
-    - [ ] ReCTS
+    - [x] ReCTS
     - [ ] SROIE
     - [ ] ArT	
     - [ ] LSVT

diff --git a/convert/RcCTS2json.py b/convert/RcCTS2json.py
@@ -0,0 +1,75 @@
+# -*- coding: utf-8 -*-
+# @Time    : 2020/3/18 14:12
+# @Author  : zhoujun
+"""
+将icdar2015数据集转换为统一格式
+"""
+import pathlib
+import numpy as np
+from tqdm import tqdm
+from convert.utils import load, save, get_file_list
+
+
+def decode_chars(char_list):
+    polygon_list = []
+    illegibility_list = []
+    text_list = []
+    for char_dict in char_list:
+        polygon_list.append(np.array(char_dict['points']).reshape(-1, 2).tolist())
+        illegibility_list.append(char_dict['ignore'])
+        text_list.append(char_dict['transcription'])
+    return polygon_list, illegibility_list, text_list
+
+
+def cvt(gt_path, save_path, img_folder):
+    """
+    将icdar2015格式的gt转换为json格式
+    :param gt_path:
+    :param save_path:
+    :return:
+    """
+    gt_dict = {'data_root': img_folder}
+    data_list = []
+    for file_path in tqdm(get_file_list(gt_path, p_postfix=['.json'])):
+        content = load(file_path)
+        file_path = pathlib.Path(file_path)
+        img_name = file_path.stem + '.jpg'
+        cur_gt = {'img_name': img_name, 'annotations': []}
+        char_polygon_list, char_illegibility_list, char_text_list = decode_chars(content['chars'])
+        for line in content['lines']:
+            cur_line_gt = {'polygon': [], 'text': '', 'illegibility': False, 'language': 'Latin'}
+            chars_gt = [{'polygon': [], 'char': '', 'illegibility': False, 'language': 'Latin'}]
+            cur_line_gt['chars'] = chars_gt
+            # 字符串级别的信息
+            cur_line_gt['polygon'] = np.array(line['points']).reshape(-1, 2).tolist()
+            cur_line_gt['text'] = line['transcription']
+            cur_line_gt['illegibility'] = True if line['ignore'] == 1 else False
+            str_len = len(line['transcription'])
+            # 字符信息
+            flag = False
+            for char_idx in range(len(char_polygon_list)):
+                for str_idx in range(1, str_len + 1):
+                    if ''.join(char_text_list[char_idx:char_idx + str_idx]) == line['transcription']:
+                        chars_gt = []
+                        for j in range(char_idx, char_idx + str_idx):
+                            chars_gt.append({'polygon': char_polygon_list[j], 'char': char_text_list[j],
+                                             'illegibility': char_illegibility_list[j], 'language': 'Latin'})
+                        cur_line_gt['chars'] = chars_gt
+                        char_polygon_list = char_polygon_list[char_idx + str_len:]
+                        char_text_list = char_text_list[char_idx + str_len:]
+                        char_illegibility_list = char_illegibility_list[char_idx + str_len:]
+                        flag = True
+                        break
+                if flag:
+                    break
+            cur_gt['annotations'].append(cur_line_gt)
+        data_list.append(cur_gt)
+    gt_dict['data_list'] = data_list
+    save(gt_dict, save_path)
+
+
+if __name__ == '__main__':
+    gt_path = r'D:\dataset\ReCTS\detection\gt'
+    img_folder = r'D:\dataset\ReCTS\detection\img'
+    save_path = r'D:\dataset\ReCTS\detection\train.json'
+    cvt(gt_path, save_path, img_folder)
diff --git a/convert/check_json.py b/convert/check_json.py
@@ -10,7 +10,7 @@
 from convert.utils import show_bbox_on_image, load_gt
 
 if __name__ == '__main__':
-    json_path = r'D:\dataset\mlt2019\detection\gt.json'
+    json_path = r'D:\dataset\ReCTS\detection\train.json'
     data = load_gt(json_path)
     for img_path, gt in data.items():
         img = Image.open(img_path)

diff --git a/convert/convert2jpg.py b/convert/convert2jpg.py
@@ -14,4 +14,5 @@
     for img_path in tqdm(get_file_list(img_folder, p_postfix=['.*'])):
         img_path = pathlib.Path(img_path)
         save_path = img_path.parent / (img_path.stem + '.jpg')
-        os.rename(img_path, save_path)
+        if img_path != save_path:
+            os.rename(img_path, save_path)
diff --git a/convert/crop_rec.py b/convert/crop_rec.py
@@ -66,8 +66,8 @@ def four_point_transform(image, pts):
 
 
 if __name__ == '__main__':
-    json_path = r'D:\dataset\COCO_Text\detection\train.json'
-    save_path = r'D:\dataset\COCO_Text\recognition\train'
+    json_path = r'D:\dataset\ReCTS\detection\train.json'
+    save_path = r'D:\dataset\ReCTS\recognition\train'
     gt_path = pathlib.Path(save_path).parent / 'train.txt'
     if os.path.exists(save_path):
         shutil.rmtree(save_path, ignore_errors=True)
@@ -79,7 +79,7 @@ def four_point_transform(image, pts):
         img_name = pathlib.Path(img_path).stem
         for i, (polygon, text, illegibility, language) in enumerate(
                 zip(gt['polygons'], gt['texts'], gt['illegibility_list'], gt['language_list'])):
-            if illegibility:
+            if illegibility or '###' in text or '*' in text:
                 continue
             polygon = np.array(polygon)
             roi_img_save_path = os.path.join(save_path, '{}_{}.jpg'.format(img_name, i))

diff --git a/convert/icdar2015tojson.py → convert/icdar20152json.py b/convert/icdar2015tojson.py → convert/icdar20152json.py
@@ -9,14 +9,14 @@
 from convert.utils import load, save, get_file_list
 
 
-def cvt(gt_path, save_path):
+def cvt(gt_path, save_path, img_folder):
     """
     将icdar2015格式的gt转换为json格式
     :param gt_path:
     :param save_path:
     :return:
     """
-    gt_dict = {'data_root': gt_path}
+    gt_dict = {'data_root': img_folder}
     data_list = []
     for file_path in tqdm(get_file_list(gt_path, p_postfix=['.txt'])):
         content = load(file_path)
@@ -41,5 +41,6 @@ def cvt(gt_path, save_path):
 
 if __name__ == '__main__':
     gt_path = r'D:\dataset\icdar2015\detection\test\gt'
+    img_folder = r'D:\dataset\icdar2015\detection\test\imgs'
     save_path = r'D:\dataset\icdar2015\detection\test.json'
-    cvt(gt_path, save_path)
+    cvt(gt_path, save_path, img_folder)
diff --git a/convert/mlt2019tojson.py → convert/mlt20192json.py b/convert/mlt2019tojson.py → convert/mlt20192json.py
@@ -10,14 +10,14 @@
 from convert.utils import load, save, get_file_list
 
 
-def cvt(gt_path, save_path):
+def cvt(gt_path, save_path, img_folder):
     """
     将icdar2015格式的gt转换为json格式
     :param gt_path:
     :param save_path:
     :return:
     """
-    gt_dict = {'data_root': gt_path}
+    gt_dict = {'data_root': img_folder}
     data_list = []
     for file_path in tqdm(get_file_list(gt_path, p_postfix=['.txt'])):
         content = load(file_path)
@@ -45,5 +45,6 @@ def cvt(gt_path, save_path):
 
 if __name__ == '__main__':
     gt_path = r'D:\dataset\mlt2019\detection\gt'
+    img_folder = r'D:\dataset\mlt2019\detection\imgs'
     save_path = r'D:\dataset\mlt2019\detection\gt.json'
-    cvt(gt_path, save_path)
+    cvt(gt_path, save_path, img_folder)
diff --git a/convert/utils.py b/convert/utils.py
@@ -126,7 +126,7 @@ def load_gt(json_path):
                 polygons.append(char_annotation['polygon'])
                 texts.append(char_annotation['char'])
                 illegibility_list.append(char_annotation['illegibility'])
-                language_list.append(char_annotation['language_'])
+                language_list.append(char_annotation['language'])
         d[img_path] = {'polygons': polygons, 'texts': texts, 'illegibility_list': illegibility_list,
                        'language_list': language_list}
-    return d
+    return d