IAM ReCTS

datasets/process/process_ESTVQA.py

@@ -5,8 +5,8 @@ def has_chinese_characters(string):
     pattern = re.compile(r'[\u4e00-\u9fa5]')
     return bool(pattern.search(string))
 if __name__ == "__main__":
-    ann_file = "/home/zhangli/GPT4/MutimodelOCR/data/ESTVQA/annotations/train.json"
-    img_file = "/home/zhangli/GPT4/MutimodelOCR/data/ESTVQA/images/train"
+    ann_file = "/home/zhangli/OCRData/data/TextVQA/ESTVQA/annotations/train.json"
+    #img_file = "/home/zhangli/GPT4/MutimodelOCR/data/ESTVQA/images/train"
     cn_list = []
     en_list= []
     with open(ann_file,'r') as f:

datasets/process/process_ReCTS.py

@@ -0,0 +1,23 @@
+import re
+import os
+def has_chinese_characters(string):
+    pattern = re.compile(r'[\u4e00-\u9fa5]')
+    return bool(pattern.search(string))
+def is_all_chinese(text):
+    """
+    判断一个字符串是否仅仅包含中文
+    """
+    pattern = re.compile(r'^[\u4e00-\u9fa5]+$')
+    return pattern.match(text) is not None
+if __name__ =='__main__':
+    file_path = "/home/zhangli/GPT4/MutimodelOCR/data/ReCTS/annotation.txt"
+    out = open("/home/zhangli/GPT4/MutimodelOCR/data/ReCTS/ann.txt",'w')
+    with open(file_path, 'r') as file:
+        data = file.readlines()
+        for line in data:
+            text = line.strip().split()[1]
+            path = os.path.join("/home/zhangli/GPT4/MutimodelOCR/data/ReCTS/crops",line.strip().split()[0])
+            if is_all_chinese(text) and os.path.exists(path):
+                out.write(line.strip())
+                out.write('\n')
+    out.close()