fix bugs

configs/rec/multi_language/rec_multi_language_lite_train.yml

@@ -17,7 +17,7 @@ Global:
   # for data or label process
   character_dict_path: 
   # Set the language of training, if set, select the default dictionary file
-  character_type: 
+  character_type: ch
   max_text_length: 25
   infer_mode: False
   use_space_char: True

ppocr/postprocess/rec_postprocess.py

@@ -24,7 +24,9 @@ class BaseRecLabelDecode(object):
                  character_type='ch',
                  use_space_char=False):
         support_character_type = [
-            'ch', 'en', 'en_sensitive', 'french', 'german', 'japan', 'korean'
+            'ch', 'en', 'en_sensitive', 'french', 'german', 'japan', 'korean', 'it',
+            'xi', 'pu', 'ru', 'ar', 'ta', 'ug', 'fa', 'ur', 'rs', 'oc', 'rsc', 'bg',
+            'uk', 'be', 'te', 'ka', 'chinese_cht', 'hi', 'mr', 'ne'
         ]
         assert character_type in support_character_type, "Only {} are supported now but get {}".format(
             support_character_type, character_type)