vsr v1.0.0

backend/ppocr/utils/__init__.py

@@ -0,0 +1,13 @@
+# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.

backend/ppocr/utils/dict/ar_dict.txt

@@ -0,0 +1,117 @@
+a
+r
+b
+i
+c
+_
+m
+g
+/
+1
+0
+I
+L
+S
+V
+R
+C
+2
+v
+l
+6
+3
+9
+.
+j
+p	
+ا
+ل
+م
+ر
+ج
+و
+ح
+ي
+ة
+5
+8
+7
+أ
+ب
+ض
+4
+ك
+س
+ه
+ث
+ن
+ط
+ع
+ت
+غ
+خ
+ف
+ئ
+ز
+إ
+د
+ص
+ظ
+ذ
+ش
+ى
+ق
+ؤ
+آ
+ء
+s
+e
+n
+w
+t
+u
+z
+d
+A
+N
+G
+h
+o
+E
+T
+H
+O
+B
+y
+F
+U
+J
+X
+W
+P
+Z
+M
+k
+q
+Y
+Q
+D
+f
+K
+x
+'
+%
+-
+#
+@
+!
+&
+$
+,
+:
+é
+?
++
+É
+(
+ 

backend/ppocr/utils/dict/arabic_dict.txt

@@ -0,0 +1,161 @@
+!
+#
+$
+%
+&
+'
+(
++
+,
+-
+.
+/
+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
+É
+é
+ء
+آ
+أ
+ؤ
+إ
+ئ
+ا
+ب
+ة
+ت
+ث
+ج
+ح
+خ
+د
+ذ
+ر
+ز
+س
+ش
+ص
+ض
+ط
+ظ
+ع
+غ
+ف
+ق
+ك
+ل
+م
+ن
+ه
+و
+ى
+ي
+ً
+ٌ
+ٍ
+َ
+ُ
+ِ
+ّ
+ْ
+ٓ
+ٔ
+ٰ
+ٱ
+ٹ
+پ
+چ
+ڈ
+ڑ
+ژ
+ک
+ڭ
+گ
+ں
+ھ
+ۀ
+ہ
+ۂ
+ۃ
+ۆ
+ۇ
+ۈ
+ۋ
+ی
+ې
+ے
+ۓ
+ە
+١
+٢
+٣
+٤
+٥
+٦
+٧
+٨
+٩

backend/ppocr/utils/dict/be_dict.txt

@@ -0,0 +1,145 @@
+b
+e
+_
+i
+m
+g
+/
+2
+0
+I
+L
+S
+V
+R
+C
+1
+v
+a
+l
+6
+9
+4
+3
+.
+j
+p	
+п
+а
+з
+б
+у
+г
+н
+ц
+ь
+8
+м
+л
+і
+о
+ў
+ы
+7
+5
+М
+х
+с
+р
+ф
+я
+е
+д
+ж
+ю
+ч
+й
+к
+Д
+в
+Б
+т
+І
+ш
+ё
+э
+К
+Л
+Н
+А
+Ж
+Г
+В
+П
+З
+Е
+О
+Р
+С
+У
+Ё
+Й
+Т
+Ч
+Э
+Ц
+Ю
+Ш
+Ф
+Х
+Я
+Ь
+Ы
+Ў
+s
+c
+n
+w
+M
+o
+t
+T
+E
+A
+B
+u
+h
+y
+k
+r
+H
+d
+Y
+O
+U
+F
+f
+x
+D
+G
+N
+K
+P
+z
+J
+X
+W
+Z
+Q
+%
+-
+q
+@
+'
+!
+#
+&
+,
+:
+$
+(
+?
+é
++
+É
+ 

backend/ppocr/utils/dict/bg_dict.txt

@@ -0,0 +1,140 @@
+!
+#
+$
+%
+&
+'
+(
++
+,
+-
+.
+/
+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
+É
+é
+А
+Б
+В
+Г
+Д
+Е
+Ж
+З
+И
+Й
+К
+Л
+М
+Н
+О
+П
+Р
+С
+Т
+У
+Ф
+Х
+Ц
+Ч
+Ш
+Щ
+Ъ
+Ю
+Я
+а
+б
+в
+г
+д
+е
+ж
+з
+и
+й
+к
+л
+м
+н
+о
+п
+р
+с
+т
+у
+ф
+х
+ц
+ч
+ш
+щ
+ъ
+ь
+ю
+я
+ 

backend/ppocr/utils/dict/cyrillic_dict.txt

@@ -0,0 +1,163 @@
+ 
+!
+#
+$
+%
+&
+'
+(
++
+,
+-
+.
+/
+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
+É
+é
+Ё
+Є
+І
+Ј
+Љ
+Ў
+А
+Б
+В
+Г
+Д
+Е
+Ж
+З
+И
+Й
+К
+Л
+М
+Н
+О
+П
+Р
+С
+Т
+У
+Ф
+Х
+Ц
+Ч
+Ш
+Щ
+Ъ
+Ы
+Ь
+Э
+Ю
+Я
+а
+б
+в
+г
+д
+е
+ж
+з
+и
+й
+к
+л
+м
+н
+о
+п
+р
+с
+т
+у
+ф
+х
+ц
+ч
+ш
+щ
+ъ
+ы
+ь
+э
+ю
+я
+ё
+ђ
+є
+і
+ј
+љ
+њ
+ћ
+ў
+џ
+Ґ
+ґ

backend/ppocr/utils/dict/devanagari_dict.txt

@@ -0,0 +1,167 @@
+ 
+!
+#
+$
+%
+&
+'
+(
++
+,
+-
+.
+/
+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
+É
+é
+ँ
+ं
+ः
+अ
+आ
+इ
+ई
+उ
+ऊ
+ऋ
+ए
+ऐ
+ऑ
+ओ
+औ
+क
+ख
+ग
+घ
+ङ
+च
+छ
+ज
+झ
+ञ
+ट
+ठ
+ड
+ढ
+ण
+त
+थ
+द
+ध
+न
+ऩ
+प
+फ
+ब
+भ
+म
+य
+र
+ऱ
+ल
+ळ
+व
+श
+ष
+स
+ह
+़
+ा
+ि
+ी
+ु
+ू
+ृ
+ॅ
+े
+ै
+ॉ
+ो
+ौ
+्
+॒
+क़
+ख़
+ग़
+ज़
+ड़
+ढ़
+फ़
+ॠ
+।
+०
+१
+२
+३
+४
+५
+६
+७
+८
+९
+॰

backend/ppocr/utils/dict/en_dict.txt

@@ -0,0 +1,95 @@
+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
+{
+|
+}
+~
+!
+"
+#
+$
+%
+&
+'
+(
+)
+*
++
+,
+-
+.
+/
+ 

backend/ppocr/utils/dict/es_dict.txt

@@ -0,0 +1,110 @@
+x
+i
+_
+m
+g
+/
+1
+0
+I
+L
+S
+V
+R
+C
+2
+v
+a
+l
+3
+6
+4
+5
+.
+j
+p
+	
+Q
+u
+e
+r
+o
+8
+7
+n
+c
+9
+t
+b
+é
+q
+d
+ó
+y
+F
+s
+,
+O
+í
+T
+f
+"
+U
+M
+h
+:
+P
+H
+A
+E
+D
+z
+N
+á
+ñ
+ú
+%
+;
+è
++
+Y
+-
+B
+G
+(
+)
+¿
+?
+w
+¡
+!
+X
+É
+K
+k
+Á
+ü
+Ú
+«
+»
+J
+'
+ö
+W
+Z
+º
+Ö
+­
+[
+]
+Ç
+ç
+à
+ä
+û
+ò
+Í
+ê
+ô
+ø
+ª

backend/ppocr/utils/dict/fa_dict.txt

@@ -0,0 +1,136 @@
+f
+a
+_
+i
+m
+g
+/
+1
+3
+I
+L
+S
+V
+R
+C
+2
+0
+v
+l
+6
+8
+5
+.
+j
+p	
+و
+د
+ر
+ك
+ن
+ش
+ه
+ا
+4
+9
+ی
+ج
+ِ
+7
+غ
+ل
+س
+ز
+ّ
+ت
+ک
+گ
+ي
+م
+ب
+ف
+چ
+خ
+ق
+ژ
+آ
+ص
+پ
+َ
+ع
+ئ
+ح
+ٔ
+ض
+ُ
+ذ
+أ
+ى
+ط
+ظ
+ث
+ة
+ً
+ء
+ؤ
+ْ
+ۀ
+إ
+ٍ
+ٌ
+ٰ
+ٓ
+ٱ
+s
+c
+e
+n
+w
+N
+E
+W
+Y
+D
+O
+H
+A
+d
+z
+r
+T
+G
+o
+t
+x
+h
+b
+B
+M
+Z
+u
+P
+F
+y
+q
+U
+K
+k
+J
+Q
+'
+X
+#
+?
+%
+$
+,
+:
+&
+!
+-
+(
+É
+@
+é
++
+ 

backend/ppocr/utils/dict/french_dict.txt

@@ -0,0 +1,136 @@
+f
+e
+n
+c
+h
+_
+i
+m
+g
+/
+r
+v
+a
+l
+t
+w
+o
+d
+6
+1
+.
+p
+B
+u
+2
+à
+3
+R
+y
+4
+U
+E
+A
+5
+P
+O
+S
+T
+D
+7
+Z
+8
+I
+N
+L
+G
+M
+H
+0
+J
+K
+-
+9
+F
+C
+V
+é
+X
+'
+s
+Q
+:
+è
+x
+b
+Y
+Œ
+É
+z
+W
+Ç
+È
+k
+Ô
+ô
+€
+À
+Ê
+q
+ù
+°
+ê
+î
+*
+Â
+j
+"
+,
+â
+%
+û
+ç
+ü
+?
+!
+;
+ö
+(
+)
+ï
+º
+ó
+ø
+å
++
+™
+á
+Ë
+<
+²
+Á
+Î
+&
+@
+œ
+ε
+Ü
+ë
+[
+]
+í
+ò
+Ö
+ä
+ß
+«
+»
+ú
+ñ
+æ
+µ
+³
+Å
+$
+#
+ 

backend/ppocr/utils/dict/german_dict.txt

@@ -0,0 +1,143 @@
+ 
+!
+"
+#
+$
+%
+&
+'
+(
+)
+*
++
+,
+-
+.
+/
+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
+£
+§
+­
+°
+´
+µ
+·
+º
+¿
+Á
+Ä
+Å
+É
+Ï
+Ô
+Ö
+Ü
+ß
+à
+á
+â
+ã
+ä
+å
+æ
+ç
+è
+é
+ê
+ë
+í
+ï
+ñ
+ò
+ó
+ô
+ö
+ø
+ù
+ú
+û
+ü
+ō
+Š
+Ÿ
+ʒ
+β
+δ
+з
+Ṡ
+‘
+€
+©
+ª
+«
+¬

backend/ppocr/utils/dict/hi_dict.txt

@@ -0,0 +1,162 @@
+ 
+!
+#
+$
+%
+&
+'
+(
++
+,
+-
+.
+/
+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
+É
+é
+ँ
+ं
+ः
+अ
+आ
+इ
+ई
+उ
+ऊ
+ऋ
+ए
+ऐ
+ऑ
+ओ
+औ
+क
+ख
+ग
+घ
+ङ
+च
+छ
+ज
+झ
+ञ
+ट
+ठ
+ड
+ढ
+ण
+त
+थ
+द
+ध
+न
+प
+फ
+ब
+भ
+म
+य
+र
+ल
+ळ
+व
+श
+ष
+स
+ह
+़
+ा
+ि
+ी
+ु
+ू
+ृ
+ॅ
+े
+ै
+ॉ
+ो
+ौ
+्
+क़
+ख़
+ग़
+ज़
+ड़
+ढ़
+फ़
+०
+१
+२
+३
+४
+५
+६
+७
+८
+९
+॰

backend/ppocr/utils/dict/it_dict.txt

@@ -0,0 +1,118 @@
+i
+t
+_
+m
+g
+/
+5
+I
+L
+S
+V
+R
+C
+2
+0
+1
+v
+a
+l
+7
+8
+9
+6
+.
+j
+p
+	
+e
+r
+o
+d
+s
+n
+3
+4
+P
+u
+c
+A
+-
+,
+"
+z
+h
+f
+b
+q
+ì
+'
+à
+O
+è
+G
+ù
+é
+ò
+;
+F
+E
+B
+N
+H
+k
+:
+U
+T
+X
+D
+K
+?
+[
+M
+­
+x
+y
+(
+)
+W
+ö
+º
+w
+]
+Q
+J
++
+ü
+!
+È
+á
+%
+=
+»
+ñ
+Ö
+Y
+ä
+í
+Z
+«
+@
+ó
+ø
+ï
+ú
+ê
+ç
+Á
+É
+Å
+ß
+{
+}
+&
+`
+û
+î
+#
+$

backend/ppocr/utils/dict/ka_dict.txt

@@ -0,0 +1,153 @@
+k
+a
+_
+i
+m
+g
+/
+1
+2
+I
+L
+S
+V
+R
+C
+0
+v
+l
+6
+4
+8
+.
+j
+p
+ಗ
+ು
+ಣ
+ಪ
+ಡ
+ಿ
+ಸ
+ಲ
+ಾ
+ದ
+್
+7
+5
+3
+ವ
+ಷ
+ಬ
+ಹ
+ೆ
+9
+ಅ
+ಳ
+ನ
+ರ
+ಉ
+ಕ
+ಎ
+ೇ
+ಂ
+ೈ
+ೊ
+ೀ
+ಯ
+ೋ
+ತ
+ಶ
+ಭ
+ಧ
+ಚ
+ಜ
+ೂ
+ಮ
+ಒ
+ೃ
+ಥ
+ಇ
+ಟ
+ಖ
+ಆ
+ಞ
+ಫ
+-
+ಢ
+ಊ
+ಓ
+ಐ
+ಃ
+ಘ
+ಝ
+ೌ
+ಠ
+ಛ
+ಔ
+ಏ
+ಈ
+ಋ
+೨
+೦
+೧
+೮
+೯
+೪
+,
+೫
+೭
+೩
+೬
+ಙ
+s
+c
+e
+n
+w
+o
+u
+t
+d
+E
+A
+T
+B
+Z
+N
+G
+O
+q
+z
+r
+x
+P
+K
+M
+J
+U
+D
+f
+F
+h
+b
+W
+Y
+y
+H
+X
+Q
+'
+#
+&
+!
+@
+$
+:
+%
+é
+É
+(
+?
++
+ 

backend/ppocr/utils/dict/kie_dict/xfund_class_list.txt

@@ -0,0 +1,4 @@
+OTHER
+QUESTION
+ANSWER
+HEADER

backend/ppocr/utils/dict/kn_dict.txt

@@ -0,0 +1,153 @@
+k
+a
+_
+i
+m
+g
+/
+1
+2
+I
+L
+S
+V
+R
+C
+0
+v
+l
+6
+4
+8
+.
+j
+p	
+ಗ
+ು
+ಣ
+ಪ
+ಡ
+ಿ
+ಸ
+ಲ
+ಾ
+ದ
+್
+7
+5
+3
+ವ
+ಷ
+ಬ
+ಹ
+ೆ
+9
+ಅ
+ಳ
+ನ
+ರ
+ಉ
+ಕ
+ಎ
+ೇ
+ಂ
+ೈ
+ೊ
+ೀ
+ಯ
+ೋ
+ತ
+ಶ
+ಭ
+ಧ
+ಚ
+ಜ
+ೂ
+ಮ
+ಒ
+ೃ
+ಥ
+ಇ
+ಟ
+ಖ
+ಆ
+ಞ
+ಫ
+-
+ಢ
+ಊ
+ಓ
+ಐ
+ಃ
+ಘ
+ಝ
+ೌ
+ಠ
+ಛ
+ಔ
+ಏ
+ಈ
+ಋ
+೨
+೦
+೧
+೮
+೯
+೪
+,
+೫
+೭
+೩
+೬
+ಙ
+s
+c
+e
+n
+w
+o
+u
+t
+d
+E
+A
+T
+B
+Z
+N
+G
+O
+q
+z
+r
+x
+P
+K
+M
+J
+U
+D
+f
+F
+h
+b
+W
+Y
+y
+H
+X
+Q
+'
+#
+&
+!
+@
+$
+:
+%
+é
+É
+(
+?
++
+ 

backend/ppocr/utils/dict/latin_dict.txt

@@ -0,0 +1,185 @@
+ 
+!
+"
+#
+$
+%
+&
+'
+(
+)
+*
++
+,
+-
+.
+/
+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
+{
+}
+¡
+£
+§
+ª
+«
+­
+°
+²
+³
+´
+µ
+·
+º
+»
+¿
+À
+Á
+Â
+Ä
+Å
+Ç
+È
+É
+Ê
+Ë
+Ì
+Í
+Î
+Ï
+Ò
+Ó
+Ô
+Õ
+Ö
+Ú
+Ü
+Ý
+ß
+à
+á
+â
+ã
+ä
+å
+æ
+ç
+è
+é
+ê
+ë
+ì
+í
+î
+ï
+ñ
+ò
+ó
+ô
+õ
+ö
+ø
+ù
+ú
+û
+ü
+ý
+ą
+Ć
+ć
+Č
+č
+Đ
+đ
+ę
+ı
+Ł
+ł
+ō
+Œ
+œ
+Š
+š
+Ÿ
+Ž
+ž
+ʒ
+β
+δ
+ε
+з
+Ṡ
+‘
+€
+™

backend/ppocr/utils/dict/layout_dict/layout_cdla_dict.txt

@@ -0,0 +1,10 @@
+text
+title
+figure
+figure_caption
+table
+table_caption
+header
+footer
+reference
+equation

backend/ppocr/utils/dict/layout_dict/layout_publaynet_dict.txt

@@ -0,0 +1,5 @@
+text
+title
+list
+table
+figure

backend/ppocr/utils/dict/layout_dict/layout_table_dict.txt

@@ -0,0 +1 @@
+table

backend/ppocr/utils/dict/mr_dict.txt

@@ -0,0 +1,153 @@
+ 
+!
+#
+$
+%
+&
+'
+(
++
+,
+-
+.
+/
+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
+É
+é
+ँ
+ं
+ः
+अ
+आ
+इ
+ई
+उ
+ऊ
+ए
+ऐ
+ऑ
+ओ
+औ
+क
+ख
+ग
+घ
+च
+छ
+ज
+झ
+ञ
+ट
+ठ
+ड
+ढ
+ण
+त
+थ
+द
+ध
+न
+प
+फ
+ब
+भ
+म
+य
+र
+ऱ
+ल
+ळ
+व
+श
+ष
+स
+ह
+़
+ा
+ि
+ी
+ु
+ू
+ृ
+ॅ
+े
+ै
+ॉ
+ो
+ौ
+्
+०
+१
+२
+३
+४
+५
+६
+७
+८
+९

backend/ppocr/utils/dict/ne_dict.txt

@@ -0,0 +1,153 @@
+ 
+!
+#
+$
+%
+&
+'
+(
++
+,
+-
+.
+/
+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
+É
+é
+ः
+अ
+आ
+इ
+ई
+उ
+ऊ
+ऋ
+ए
+ऐ
+ओ
+औ
+क
+ख
+ग
+घ
+ङ
+च
+छ
+ज
+झ
+ञ
+ट
+ठ
+ड
+ढ
+ण
+त
+थ
+द
+ध
+न
+ऩ
+प
+फ
+ब
+भ
+म
+य
+र
+ऱ
+ल
+व
+श
+ष
+स
+ह
+़
+ा
+ि
+ी
+ु
+ू
+ृ
+े
+ै
+ो
+ौ
+्
+॒
+ॠ
+।
+०
+१
+२
+३
+४
+५
+६
+७
+८
+९

backend/ppocr/utils/dict/oc_dict.txt

@@ -0,0 +1,96 @@
+o
+c
+_
+i
+m
+g
+/
+2
+0
+I
+L
+S
+V
+R
+C
+1
+v
+a
+l
+4
+3
+.
+j
+p	
+r
+e
+è
+t
+9
+7
+5
+8
+n
+'
+b
+s
+6
+q
+u
+á
+d
+ò
+à
+h
+z
+f
+ï
+í
+A
+ç
+x
+ó
+é
+P
+O
+Ò
+ü
+k
+À
+F
+-
+ú
+­
+æ
+Á
+D
+E
+w
+K
+T
+N
+y
+U
+Z
+G
+B
+J
+H
+M
+W
+Y
+X
+Q
+%
+$
+,
+@
+&
+!
+:
+(
+#
+?
++
+É
+ 

backend/ppocr/utils/dict/pt_dict.txt

@@ -0,0 +1,130 @@
+p
+u
+_
+i
+m
+g
+/
+8
+I
+L
+S
+V
+R
+C
+2
+0
+1
+v
+a
+l
+6
+7
+4
+5
+.
+j
+	
+q
+e
+s
+t
+ã
+o
+x
+9
+c
+n
+r
+z
+ç
+õ
+3
+A
+U
+d
+º
+ô
+­
+,
+E
+;
+ó
+á
+b
+D
+?
+ú
+ê
+-
+h
+P
+f
+à
+N
+í
+O
+M
+G
+É
+é
+â
+F
+:
+T
+Á
+"
+Q
+)
+W
+J
+B
+H
+(
+ö
+%
+Ö
+«
+w
+K
+y
+!
+k
+]
+'
+Z
++
+Ç
+Õ
+Y
+À
+X
+µ
+»
+ª
+Í
+ü
+ä
+´
+è
+ñ
+ß
+ï
+Ú
+ë
+Ô
+Ï
+Ó
+[
+Ì
+<
+Â
+ò
+§
+³
+ø
+å
+#
+$
+&
+@

backend/ppocr/utils/dict/pu_dict.txt

@@ -0,0 +1,130 @@
+p
+u
+_
+i
+m
+g
+/
+8
+I
+L
+S
+V
+R
+C
+2
+0
+1
+v
+a
+l
+6
+7
+4
+5
+.
+j
+	
+q
+e
+s
+t
+ã
+o
+x
+9
+c
+n
+r
+z
+ç
+õ
+3
+A
+U
+d
+º
+ô
+­
+,
+E
+;
+ó
+á
+b
+D
+?
+ú
+ê
+-
+h
+P
+f
+à
+N
+í
+O
+M
+G
+É
+é
+â
+F
+:
+T
+Á
+"
+Q
+)
+W
+J
+B
+H
+(
+ö
+%
+Ö
+«
+w
+K
+y
+!
+k
+]
+'
+Z
++
+Ç
+Õ
+Y
+À
+X
+µ
+»
+ª
+Í
+ü
+ä
+´
+è
+ñ
+ß
+ï
+Ú
+ë
+Ô
+Ï
+Ó
+[
+Ì
+<
+Â
+ò
+§
+³
+ø
+å
+#
+$
+&
+@

backend/ppocr/utils/dict/rs_cyrillic_dict.txt

@@ -0,0 +1,134 @@
+r
+s
+c
+_
+i
+m
+g
+/
+5
+I
+L
+S
+V
+R
+C
+2
+0
+1
+v
+a
+l
+9
+7
+8
+.
+j
+p	
+м
+а
+с
+и
+р
+ћ
+е
+ш
+3
+4
+о
+г
+н
+з
+в
+л
+6
+т
+ж
+у
+к
+п
+њ
+д
+ч
+С
+ј
+ф
+ц
+љ
+х
+О
+И
+А
+б
+Ш
+К
+ђ
+џ
+М
+В
+З
+Д
+Р
+У
+Н
+Т
+Б
+?
+П
+Х
+Ј
+Ц
+Г
+Љ
+Л
+Ф
+e
+n
+w
+E
+F
+A
+N
+f
+o
+b
+M
+G
+t
+y
+W
+k
+P
+u
+H
+B
+T
+z
+h
+O
+Y
+d
+U
+K
+D
+x
+X
+J
+Z
+Q
+q
+'
+-
+@
+é
+#
+!
+,
+%
+$
+:
+&
++
+(
+É
+ 

backend/ppocr/utils/dict/rs_dict.txt

@@ -0,0 +1,91 @@
+r
+s
+_
+i
+m
+g
+/
+1
+I
+L
+S
+V
+R
+C
+2
+0
+v
+a
+l
+7
+5
+8
+6
+.
+j
+p
+	
+t
+d
+9
+3
+e
+š
+4
+k
+u
+ć
+c
+n
+đ
+o
+z
+č
+b
+ž
+f
+Z
+T
+h
+M
+F
+O
+Š
+B
+H
+A
+E
+Đ
+Ž
+D
+P
+G
+Č
+K
+U
+N
+J
+Ć
+w
+y
+W
+x
+Y
+X
+q
+Q
+#
+&
+$
+,
+-
+%
+'
+@
+!
+:
+?
+(
+É
+é
++

backend/ppocr/utils/dict/rs_latin_dict.txt

@@ -0,0 +1,91 @@
+r
+s
+_
+i
+m
+g
+/
+1
+I
+L
+S
+V
+R
+C
+2
+0
+v
+a
+l
+7
+5
+8
+6
+.
+j
+p
+	
+t
+d
+9
+3
+e
+š
+4
+k
+u
+ć
+c
+n
+đ
+o
+z
+č
+b
+ž
+f
+Z
+T
+h
+M
+F
+O
+Š
+B
+H
+A
+E
+Đ
+Ž
+D
+P
+G
+Č
+K
+U
+N
+J
+Ć
+w
+y
+W
+x
+Y
+X
+q
+Q
+#
+&
+$
+,
+-
+%
+'
+@
+!
+:
+?
+(
+É
+é
++

backend/ppocr/utils/dict/rsc_dict.txt

@@ -0,0 +1,134 @@
+r
+s
+c
+_
+i
+m
+g
+/
+5
+I
+L
+S
+V
+R
+C
+2
+0
+1
+v
+a
+l
+9
+7
+8
+.
+j
+p	
+м
+а
+с
+и
+р
+ћ
+е
+ш
+3
+4
+о
+г
+н
+з
+в
+л
+6
+т
+ж
+у
+к
+п
+њ
+д
+ч
+С
+ј
+ф
+ц
+љ
+х
+О
+И
+А
+б
+Ш
+К
+ђ
+џ
+М
+В
+З
+Д
+Р
+У
+Н
+Т
+Б
+?
+П
+Х
+Ј
+Ц
+Г
+Љ
+Л
+Ф
+e
+n
+w
+E
+F
+A
+N
+f
+o
+b
+M
+G
+t
+y
+W
+k
+P
+u
+H
+B
+T
+z
+h
+O
+Y
+d
+U
+K
+D
+x
+X
+J
+Z
+Q
+q
+'
+-
+@
+é
+#
+!
+,
+%
+$
+:
+&
++
+(
+É
+ 

backend/ppocr/utils/dict/ru_dict.txt

@@ -0,0 +1,163 @@
+
+!
+#
+$
+%
+&
+'
+(
++
+,
+-
+.
+/
+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
+É
+é
+Ё
+Є
+І
+Ј
+Љ
+Ў
+А
+Б
+В
+Г
+Д
+Е
+Ж
+З
+И
+Й
+К
+Л
+М
+Н
+О
+П
+Р
+С
+Т
+У
+Ф
+Х
+Ц
+Ч
+Ш
+Щ
+Ъ
+Ы
+Ь
+Э
+Ю
+Я
+а
+б
+в
+г
+д
+е
+ж
+з
+и
+й
+к
+л
+м
+н
+о
+п
+р
+с
+т
+у
+ф
+х
+ц
+ч
+ш
+щ
+ъ
+ы
+ь
+э
+ю
+я
+ё
+ђ
+є
+і
+ј
+љ
+њ
+ћ
+ў
+џ
+Ґ
+ґ

backend/ppocr/utils/dict/spin_dict.txt

@@ -0,0 +1,68 @@
+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
+:
+(
+'
+-
+,
+%
+>
+.
+[
+?
+)
+"
+=
+_
+*
+]
+;
+&
++
+$
+@
+/
+|
+!
+<
+#
+`
+{
+~
+\
+}
+^

backend/ppocr/utils/dict/ta_dict.txt

@@ -0,0 +1,128 @@
+t
+a
+_
+i
+m
+g
+/
+3
+I
+L
+S
+V
+R
+C
+2
+0
+1
+v
+l
+9
+7
+8
+.
+j
+p
+ப
+ூ
+த
+ம
+ி
+வ
+ர
+்
+ந
+ோ
+ன
+6
+ஆ
+ற
+ல
+5
+ள
+ா
+ொ
+ழ
+ு
+4
+ெ
+ண
+க
+ட
+ை
+ே
+ச
+ய
+ஒ
+இ
+அ
+ங
+உ
+ீ
+ஞ
+எ
+ஓ
+ஃ
+ஜ
+ஷ
+ஸ
+ஏ
+ஊ
+ஹ
+ஈ
+ஐ
+ௌ
+ஔ
+s
+c
+e
+n
+w
+F
+T
+O
+P
+K
+A
+N
+G
+Y
+E
+M
+H
+U
+B
+o
+b
+D
+d
+r
+W
+u
+y
+f
+X
+k
+q
+h
+J
+z
+Z
+Q
+x
+-
+'
+$
+,
+%
+@
+é
+!
+#
++
+É
+&
+:
+(
+?
+ 

backend/ppocr/utils/dict/table_dict.txt

@@ -0,0 +1,277 @@
+←
+</overline>
+☆
+─
+α
+
+ 
+⋅
+$
+ω
+ψ
+χ
+(
+υ
+≥
+σ
+,
+ρ
+ε
+0
+■
+4
+8
+✗
+b
+<
+✓
+Ψ
+Ω
+€
+D
+3
+Π
+H
+║
+</strike>
+L
+Φ
+Χ
+θ
+P
+κ
+λ
+μ
+T
+ξ
+X
+β
+γ
+δ
+\
+ζ
+η
+`
+d
+<strike>
+h
+f
+l
+Θ
+p
+√
+t
+</sub>
+x
+Β
+Γ
+Δ
+|
+ǂ
+ɛ
+j
+̧
+➢
+⁡
+̌
+′
+«
+△
+▲
+#
+</b>
+'
+Ι
++
+¶
+/
+▼
+⇑
+□
+·
+7
+▪
+;
+?
+➔
+∩
+C
+÷
+G
+⇒
+K
+<sup>
+O
+S
+С
+W
+Α
+[
+○
+_
+●
+‡
+c
+z
+g
+<i>
+o
+<sub>
+〈
+〉
+s
+⩽
+w
+φ
+ʹ
+{
+»
+∣
+̆
+e
+ˆ
+∈
+τ
+◆
+ι
+∅
+∆
+∙
+∘
+Ø
+ß
+✔
+∞
+∑
+−
+×
+◊
+∗
+∖
+˃
+˂
+∫
+"
+i
+&
+π
+↔
+*
+∥
+æ
+∧
+.
+⁄
+ø
+Q
+∼
+6
+⁎
+:
+★
+>
+a
+B
+≈
+F
+J
+̄
+N
+♯
+R
+V
+<overline>
+―
+Z
+♣
+^
+¤
+¥
+§
+<underline>
+¢
+£
+≦
+­
+≤
+‖
+Λ
+©
+n
+↓
+→
+↑
+r
+°
+±
+v
+<b>
+♂
+k
+♀
+~
+ᅟ
+̇
+@
+”
+♦
+ł
+®
+⊕
+„
+!
+</sup>
+%
+⇓
+)
+-
+1
+5
+9
+=
+А
+A
+‰
+⋆
+Σ
+E
+◦
+I
+※
+M
+m
+̨
+⩾
+†
+</i>
+•
+U
+Y
+

+]
+̸
+2
+‐
+–
+‒
+̂
+—
+̀
+́
+’
+‘
+⋮
+⋯
+̊
+“
+̈
+≧
+q
+u
+ı
+y
+</underline>
+​
+̃
+}
+ν

backend/ppocr/utils/dict/table_master_structure_dict.txt

@@ -0,0 +1,39 @@
+<thead>
+<tr>
+<td></td>
+</tr>
+</thead>
+<tbody>
+<eb></eb>
+</tbody>
+<td
+ colspan="5"
+>
+</td>
+ colspan="2"
+ colspan="3"
+<eb2></eb2>
+<eb1></eb1>
+ rowspan="2"
+ colspan="4"
+ colspan="6"
+ rowspan="3"
+ colspan="9"
+ colspan="10"
+ colspan="7"
+ rowspan="4"
+ rowspan="5"
+ rowspan="9"
+ colspan="8"
+ rowspan="8"
+ rowspan="6"
+ rowspan="7"
+ rowspan="10"
+<eb3></eb3>
+<eb4></eb4>
+<eb5></eb5>
+<eb6></eb6>
+<eb7></eb7>
+<eb8></eb8>
+<eb9></eb9>
+<eb10></eb10>

backend/ppocr/utils/dict/table_structure_dict.txt

@@ -0,0 +1,28 @@
+<thead>
+<tr>
+<td>
+</td>
+</tr>
+</thead>
+<tbody>
+</tbody>
+<td
+ colspan="5"
+>
+ colspan="2"
+ colspan="3"
+ rowspan="2"
+ colspan="4"
+ colspan="6"
+ rowspan="3"
+ colspan="9"
+ colspan="10"
+ colspan="7"
+ rowspan="4"
+ rowspan="5"
+ rowspan="9"
+ colspan="8"
+ rowspan="8"
+ rowspan="6"
+ rowspan="7"
+ rowspan="10"

backend/ppocr/utils/dict/table_structure_dict_ch.txt

@@ -0,0 +1,48 @@
+<thead>
+</thead>
+<tbody>
+</tbody>
+<tr>
+</tr>
+<td>
+<td
+>
+</td>
+ colspan="2"
+ colspan="3"
+ colspan="4"
+ colspan="5"
+ colspan="6"
+ colspan="7"
+ colspan="8"
+ colspan="9"
+ colspan="10"
+ colspan="11"
+ colspan="12"
+ colspan="13"
+ colspan="14"
+ colspan="15"
+ colspan="16"
+ colspan="17"
+ colspan="18"
+ colspan="19"
+ colspan="20"
+ rowspan="2"
+ rowspan="3"
+ rowspan="4"
+ rowspan="5"
+ rowspan="6"
+ rowspan="7"
+ rowspan="8"
+ rowspan="9"
+ rowspan="10"
+ rowspan="11"
+ rowspan="12"
+ rowspan="13"
+ rowspan="14"
+ rowspan="15"
+ rowspan="16"
+ rowspan="17"
+ rowspan="18"
+ rowspan="19"
+ rowspan="20"

backend/ppocr/utils/dict/te_dict.txt

@@ -0,0 +1,151 @@
+t
+e
+_
+i
+m
+g
+/
+5
+I
+L
+S
+V
+R
+C
+2
+0
+1
+v
+a
+l
+3
+4
+8
+9
+.
+j
+p
+త
+ె
+ర
+క
+్
+ి
+ం
+చ
+ే
+ద
+ు
+7
+6
+ఉ
+ా
+మ
+ట
+ో
+వ
+ప
+ల
+శ
+ఆ
+య
+ై
+భ
+'
+ీ
+గ
+ూ
+డ
+ధ
+హ
+న
+జ
+స
+[
+‌
+ష
+అ
+ణ
+ఫ
+బ
+ఎ
+;
+ళ
+థ
+ొ
+ఠ
+ృ
+ఒ
+ఇ
+ః
+ఊ
+ఖ
+-
+ఐ
+ఘ
+ౌ
+ఏ
+ఈ
+ఛ
+,
+ఓ
+ఞ
+|
+?
+:
+ఢ
+"
+(
+”
+!
++
+)
+*
+=
+&
+“
+€
+]
+£
+$
+s
+c
+n
+w
+k
+J
+G
+u
+d
+r
+E
+o
+h
+y
+b
+f
+B
+M
+O
+T
+N
+D
+P
+A
+F
+x
+W
+Y
+U
+H
+K
+X
+z
+Z
+Q
+q
+É
+%
+#
+@
+é

backend/ppocr/utils/dict/ug_dict.txt

@@ -0,0 +1,114 @@
+u
+g
+_
+i
+m
+/
+1
+I
+L
+S
+V
+R
+C
+2
+0
+v
+a
+l
+8
+5
+3
+6
+9
+.
+j
+p
+	
+ق
+ا
+پ
+ل
+4
+7
+ئ
+ى
+ش
+ت
+ي
+ك
+د
+ف
+ر
+و
+ن
+ب
+ە
+خ
+ې
+چ
+ۇ
+ز
+س
+م
+ۋ
+گ
+ڭ
+ۆ
+ۈ
+ج
+غ
+ھ
+ژ
+s
+c
+e
+n
+w
+P
+E
+D
+U
+d
+r
+b
+y
+B
+o
+O
+Y
+N
+T
+k
+t
+h
+A
+H
+F
+z
+W
+K
+G
+M
+f
+Z
+X
+Q
+J
+x
+q
+-
+!
+%
+#
+?
+:
+$
+,
+&
+'
+É
+@
+é
+(
++

backend/ppocr/utils/dict/uk_dict.txt

@@ -0,0 +1,142 @@
+u
+k
+_
+i
+m
+g
+/
+1
+6
+I
+L
+S
+V
+R
+C
+2
+0
+v
+a
+l
+7
+9
+.
+j
+p	
+в
+і
+д
+п
+о
+н
+с
+т
+ю
+4
+5
+3
+а
+и
+м
+е
+р
+ч
+у
+Б
+з
+л
+к
+8
+А
+В
+г
+є
+б
+ь
+х
+ґ
+ш
+ц
+ф
+я
+щ
+ж
+Г
+Х
+У
+Т
+Е
+І
+Н
+П
+З
+Л
+Ю
+С
+Д
+М
+К
+Р
+Ф
+О
+Ц
+И
+Я
+Ч
+Ш
+Ж
+Є
+Ґ
+Ь
+s
+c
+e
+n
+w
+A
+P
+r
+E
+t
+o
+h
+d
+y
+M
+G
+N
+F
+B
+T
+D
+U
+O
+W
+Z
+f
+H
+Y
+b
+K
+z
+x
+Q
+X
+q
+J
+$
+-
+'
+#
+&
+%
+?
+:
+!
+,
++
+@
+(
+é
+É
+ 

backend/ppocr/utils/dict/ur_dict.txt

@@ -0,0 +1,137 @@
+u
+r
+_
+i
+m
+g
+/
+3
+I
+L
+S
+V
+R
+C
+2
+0
+1
+v
+a
+l
+9
+7
+8
+.
+j
+p
+	
+چ
+ٹ
+پ
+ا
+ئ
+ی
+ے
+4
+6
+و
+ل
+ن
+ڈ
+ھ
+ک
+ت
+ش
+ف
+ق
+ر
+د
+5
+ب
+ج
+خ
+ہ
+س
+ز
+غ
+ڑ
+ں
+آ
+م
+ؤ
+ط
+ص
+ح
+ع
+گ
+ث
+ض
+ذ
+ۓ
+ِ
+ء
+ظ
+ً
+ي
+ُ
+ۃ
+أ
+ٰ
+ە
+ژ
+ۂ
+ة
+ّ
+ك
+ه
+s
+c
+e
+n
+w
+o
+d
+t
+D
+M
+T
+U
+E
+b
+P
+h
+y
+W
+H
+A
+x
+B
+O
+N
+G
+Y
+Q
+F
+k
+K
+q
+J
+Z
+f
+z
+X
+'
+@
+&
+!
+,
+:
+$
+-
+#
+?
+%
+é
++
+(
+É

backend/ppocr/utils/dict/xi_dict.txt

@@ -0,0 +1,110 @@
+x
+i
+_
+m
+g
+/
+1
+0
+I
+L
+S
+V
+R
+C
+2
+v
+a
+l
+3
+6
+4
+5
+.
+j
+p
+	
+Q
+u
+e
+r
+o
+8
+7
+n
+c
+9
+t
+b
+é
+q
+d
+ó
+y
+F
+s
+,
+O
+í
+T
+f
+"
+U
+M
+h
+:
+P
+H
+A
+E
+D
+z
+N
+á
+ñ
+ú
+%
+;
+è
++
+Y
+-
+B
+G
+(
+)
+¿
+?
+w
+¡
+!
+X
+É
+K
+k
+Á
+ü
+Ú
+«
+»
+J
+'
+ö
+W
+Z
+º
+Ö
+­
+[
+]
+Ç
+ç
+à
+ä
+û
+ò
+Í
+ê
+ô
+ø
+ª

backend/ppocr/utils/e2e_metric/Deteval.py

@@ -0,0 +1,574 @@
+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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 numpy as np
+import scipy.io as io
+from ppocr.utils.e2e_metric.polygon_fast import iod, area_of_intersection, area
+
+
+def get_socre_A(gt_dir, pred_dict):
+    allInputs = 1
+
+    def input_reading_mod(pred_dict):
+        """This helper reads input from txt files"""
+        det = []
+        n = len(pred_dict)
+        for i in range(n):
+            points = pred_dict[i]['points']
+            text = pred_dict[i]['texts']
+            point = ",".join(map(str, points.reshape(-1, )))
+            det.append([point, text])
+        return det
+
+    def gt_reading_mod(gt_dict):
+        """This helper reads groundtruths from mat files"""
+        gt = []
+        n = len(gt_dict)
+        for i in range(n):
+            points = gt_dict[i]['points'].tolist()
+            h = len(points)
+            text = gt_dict[i]['text']
+            xx = [
+                np.array(
+                    ['x:'], dtype='<U2'), 0, np.array(
+                        ['y:'], dtype='<U2'), 0, np.array(
+                            ['#'], dtype='<U1'), np.array(
+                                ['#'], dtype='<U1')
+            ]
+            t_x, t_y = [], []
+            for j in range(h):
+                t_x.append(points[j][0])
+                t_y.append(points[j][1])
+            xx[1] = np.array([t_x], dtype='int16')
+            xx[3] = np.array([t_y], dtype='int16')
+            if text != "":
+                xx[4] = np.array([text], dtype='U{}'.format(len(text)))
+                xx[5] = np.array(['c'], dtype='<U1')
+            gt.append(xx)
+        return gt
+
+    def detection_filtering(detections, groundtruths, threshold=0.5):
+        for gt_id, gt in enumerate(groundtruths):
+            if (gt[5] == '#') and (gt[1].shape[1] > 1):
+                gt_x = list(map(int, np.squeeze(gt[1])))
+                gt_y = list(map(int, np.squeeze(gt[3])))
+                for det_id, detection in enumerate(detections):
+                    detection_orig = detection
+                    detection = [float(x) for x in detection[0].split(',')]
+                    detection = list(map(int, detection))
+                    det_x = detection[0::2]
+                    det_y = detection[1::2]
+                    det_gt_iou = iod(det_x, det_y, gt_x, gt_y)
+                    if det_gt_iou > threshold:
+                        detections[det_id] = []
+
+                detections[:] = [item for item in detections if item != []]
+        return detections
+
+    def sigma_calculation(det_x, det_y, gt_x, gt_y):
+        """
+        sigma = inter_area / gt_area
+        """
+        return np.round((area_of_intersection(det_x, det_y, gt_x, gt_y) /
+                         area(gt_x, gt_y)), 2)
+
+    def tau_calculation(det_x, det_y, gt_x, gt_y):
+        if area(det_x, det_y) == 0.0:
+            return 0
+        return np.round((area_of_intersection(det_x, det_y, gt_x, gt_y) /
+                         area(det_x, det_y)), 2)
+
+    ##############################Initialization###################################
+    # global_sigma = []
+    # global_tau = []
+    # global_pred_str = []
+    # global_gt_str = []
+    ###############################################################################
+
+    for input_id in range(allInputs):
+        if (input_id != '.DS_Store') and (input_id != 'Pascal_result.txt') and (
+                input_id != 'Pascal_result_curved.txt') and (input_id != 'Pascal_result_non_curved.txt') and (
+                input_id != 'Deteval_result.txt') and (input_id != 'Deteval_result_curved.txt') \
+                and (input_id != 'Deteval_result_non_curved.txt'):
+            detections = input_reading_mod(pred_dict)
+            groundtruths = gt_reading_mod(gt_dir)
+            detections = detection_filtering(
+                detections,
+                groundtruths)  # filters detections overlapping with DC area
+            dc_id = []
+            for i in range(len(groundtruths)):
+                if groundtruths[i][5] == '#':
+                    dc_id.append(i)
+            cnt = 0
+            for a in dc_id:
+                num = a - cnt
+                del groundtruths[num]
+                cnt += 1
+
+            local_sigma_table = np.zeros((len(groundtruths), len(detections)))
+            local_tau_table = np.zeros((len(groundtruths), len(detections)))
+            local_pred_str = {}
+            local_gt_str = {}
+
+            for gt_id, gt in enumerate(groundtruths):
+                if len(detections) > 0:
+                    for det_id, detection in enumerate(detections):
+                        detection_orig = detection
+                        detection = [float(x) for x in detection[0].split(',')]
+                        detection = list(map(int, detection))
+                        pred_seq_str = detection_orig[1].strip()
+                        det_x = detection[0::2]
+                        det_y = detection[1::2]
+                        gt_x = list(map(int, np.squeeze(gt[1])))
+                        gt_y = list(map(int, np.squeeze(gt[3])))
+                        gt_seq_str = str(gt[4].tolist()[0])
+
+                        local_sigma_table[gt_id, det_id] = sigma_calculation(
+                            det_x, det_y, gt_x, gt_y)
+                        local_tau_table[gt_id, det_id] = tau_calculation(
+                            det_x, det_y, gt_x, gt_y)
+                        local_pred_str[det_id] = pred_seq_str
+                        local_gt_str[gt_id] = gt_seq_str
+
+            global_sigma = local_sigma_table
+            global_tau = local_tau_table
+            global_pred_str = local_pred_str
+            global_gt_str = local_gt_str
+
+    single_data = {}
+    single_data['sigma'] = global_sigma
+    single_data['global_tau'] = global_tau
+    single_data['global_pred_str'] = global_pred_str
+    single_data['global_gt_str'] = global_gt_str
+    return single_data
+
+
+def get_socre_B(gt_dir, img_id, pred_dict):
+    allInputs = 1
+
+    def input_reading_mod(pred_dict):
+        """This helper reads input from txt files"""
+        det = []
+        n = len(pred_dict)
+        for i in range(n):
+            points = pred_dict[i]['points']
+            text = pred_dict[i]['texts']
+            point = ",".join(map(str, points.reshape(-1, )))
+            det.append([point, text])
+        return det
+
+    def gt_reading_mod(gt_dir, gt_id):
+        gt = io.loadmat('%s/poly_gt_img%s.mat' % (gt_dir, gt_id))
+        gt = gt['polygt']
+        return gt
+
+    def detection_filtering(detections, groundtruths, threshold=0.5):
+        for gt_id, gt in enumerate(groundtruths):
+            if (gt[5] == '#') and (gt[1].shape[1] > 1):
+                gt_x = list(map(int, np.squeeze(gt[1])))
+                gt_y = list(map(int, np.squeeze(gt[3])))
+                for det_id, detection in enumerate(detections):
+                    detection_orig = detection
+                    detection = [float(x) for x in detection[0].split(',')]
+                    detection = list(map(int, detection))
+                    det_x = detection[0::2]
+                    det_y = detection[1::2]
+                    det_gt_iou = iod(det_x, det_y, gt_x, gt_y)
+                    if det_gt_iou > threshold:
+                        detections[det_id] = []
+
+                detections[:] = [item for item in detections if item != []]
+        return detections
+
+    def sigma_calculation(det_x, det_y, gt_x, gt_y):
+        """
+        sigma = inter_area / gt_area
+        """
+        return np.round((area_of_intersection(det_x, det_y, gt_x, gt_y) /
+                         area(gt_x, gt_y)), 2)
+
+    def tau_calculation(det_x, det_y, gt_x, gt_y):
+        if area(det_x, det_y) == 0.0:
+            return 0
+        return np.round((area_of_intersection(det_x, det_y, gt_x, gt_y) /
+                         area(det_x, det_y)), 2)
+
+    ##############################Initialization###################################
+    # global_sigma = []
+    # global_tau = []
+    # global_pred_str = []
+    # global_gt_str = []
+    ###############################################################################
+
+    for input_id in range(allInputs):
+        if (input_id != '.DS_Store') and (input_id != 'Pascal_result.txt') and (
+                input_id != 'Pascal_result_curved.txt') and (input_id != 'Pascal_result_non_curved.txt') and (
+                input_id != 'Deteval_result.txt') and (input_id != 'Deteval_result_curved.txt') \
+                and (input_id != 'Deteval_result_non_curved.txt'):
+            detections = input_reading_mod(pred_dict)
+            groundtruths = gt_reading_mod(gt_dir, img_id).tolist()
+            detections = detection_filtering(
+                detections,
+                groundtruths)  # filters detections overlapping with DC area
+            dc_id = []
+            for i in range(len(groundtruths)):
+                if groundtruths[i][5] == '#':
+                    dc_id.append(i)
+            cnt = 0
+            for a in dc_id:
+                num = a - cnt
+                del groundtruths[num]
+                cnt += 1
+
+            local_sigma_table = np.zeros((len(groundtruths), len(detections)))
+            local_tau_table = np.zeros((len(groundtruths), len(detections)))
+            local_pred_str = {}
+            local_gt_str = {}
+
+            for gt_id, gt in enumerate(groundtruths):
+                if len(detections) > 0:
+                    for det_id, detection in enumerate(detections):
+                        detection_orig = detection
+                        detection = [float(x) for x in detection[0].split(',')]
+                        detection = list(map(int, detection))
+                        pred_seq_str = detection_orig[1].strip()
+                        det_x = detection[0::2]
+                        det_y = detection[1::2]
+                        gt_x = list(map(int, np.squeeze(gt[1])))
+                        gt_y = list(map(int, np.squeeze(gt[3])))
+                        gt_seq_str = str(gt[4].tolist()[0])
+
+                        local_sigma_table[gt_id, det_id] = sigma_calculation(
+                            det_x, det_y, gt_x, gt_y)
+                        local_tau_table[gt_id, det_id] = tau_calculation(
+                            det_x, det_y, gt_x, gt_y)
+                        local_pred_str[det_id] = pred_seq_str
+                        local_gt_str[gt_id] = gt_seq_str
+
+            global_sigma = local_sigma_table
+            global_tau = local_tau_table
+            global_pred_str = local_pred_str
+            global_gt_str = local_gt_str
+
+    single_data = {}
+    single_data['sigma'] = global_sigma
+    single_data['global_tau'] = global_tau
+    single_data['global_pred_str'] = global_pred_str
+    single_data['global_gt_str'] = global_gt_str
+    return single_data
+
+
+def combine_results(all_data):
+    tr = 0.7
+    tp = 0.6
+    fsc_k = 0.8
+    k = 2
+    global_sigma = []
+    global_tau = []
+    global_pred_str = []
+    global_gt_str = []
+    for data in all_data:
+        global_sigma.append(data['sigma'])
+        global_tau.append(data['global_tau'])
+        global_pred_str.append(data['global_pred_str'])
+        global_gt_str.append(data['global_gt_str'])
+
+    global_accumulative_recall = 0
+    global_accumulative_precision = 0
+    total_num_gt = 0
+    total_num_det = 0
+    hit_str_count = 0
+    hit_count = 0
+
+    def one_to_one(local_sigma_table, local_tau_table,
+                   local_accumulative_recall, local_accumulative_precision,
+                   global_accumulative_recall, global_accumulative_precision,
+                   gt_flag, det_flag, idy):
+        hit_str_num = 0
+        for gt_id in range(num_gt):
+            gt_matching_qualified_sigma_candidates = np.where(
+                local_sigma_table[gt_id, :] > tr)
+            gt_matching_num_qualified_sigma_candidates = gt_matching_qualified_sigma_candidates[
+                0].shape[0]
+            gt_matching_qualified_tau_candidates = np.where(
+                local_tau_table[gt_id, :] > tp)
+            gt_matching_num_qualified_tau_candidates = gt_matching_qualified_tau_candidates[
+                0].shape[0]
+
+            det_matching_qualified_sigma_candidates = np.where(
+                local_sigma_table[:, gt_matching_qualified_sigma_candidates[0]]
+                > tr)
+            det_matching_num_qualified_sigma_candidates = det_matching_qualified_sigma_candidates[
+                0].shape[0]
+            det_matching_qualified_tau_candidates = np.where(
+                local_tau_table[:, gt_matching_qualified_tau_candidates[0]] >
+                tp)
+            det_matching_num_qualified_tau_candidates = det_matching_qualified_tau_candidates[
+                0].shape[0]
+
+            if (gt_matching_num_qualified_sigma_candidates == 1) and (gt_matching_num_qualified_tau_candidates == 1) and \
+                    (det_matching_num_qualified_sigma_candidates == 1) and (
+                    det_matching_num_qualified_tau_candidates == 1):
+                global_accumulative_recall = global_accumulative_recall + 1.0
+                global_accumulative_precision = global_accumulative_precision + 1.0
+                local_accumulative_recall = local_accumulative_recall + 1.0
+                local_accumulative_precision = local_accumulative_precision + 1.0
+
+                gt_flag[0, gt_id] = 1
+                matched_det_id = np.where(local_sigma_table[gt_id, :] > tr)
+                # recg start
+                gt_str_cur = global_gt_str[idy][gt_id]
+                pred_str_cur = global_pred_str[idy][matched_det_id[0].tolist()[
+                    0]]
+                if pred_str_cur == gt_str_cur:
+                    hit_str_num += 1
+                else:
+                    if pred_str_cur.lower() == gt_str_cur.lower():
+                        hit_str_num += 1
+                # recg end
+                det_flag[0, matched_det_id] = 1
+        return local_accumulative_recall, local_accumulative_precision, global_accumulative_recall, global_accumulative_precision, gt_flag, det_flag, hit_str_num
+
+    def one_to_many(local_sigma_table, local_tau_table,
+                    local_accumulative_recall, local_accumulative_precision,
+                    global_accumulative_recall, global_accumulative_precision,
+                    gt_flag, det_flag, idy):
+        hit_str_num = 0
+        for gt_id in range(num_gt):
+            # skip the following if the groundtruth was matched
+            if gt_flag[0, gt_id] > 0:
+                continue
+
+            non_zero_in_sigma = np.where(local_sigma_table[gt_id, :] > 0)
+            num_non_zero_in_sigma = non_zero_in_sigma[0].shape[0]
+
+            if num_non_zero_in_sigma >= k:
+                ####search for all detections that overlaps with this groundtruth
+                qualified_tau_candidates = np.where((local_tau_table[
+                    gt_id, :] >= tp) & (det_flag[0, :] == 0))
+                num_qualified_tau_candidates = qualified_tau_candidates[
+                    0].shape[0]
+
+                if num_qualified_tau_candidates == 1:
+                    if ((local_tau_table[gt_id, qualified_tau_candidates] >= tp)
+                            and
+                        (local_sigma_table[gt_id, qualified_tau_candidates] >=
+                         tr)):
+                        # became an one-to-one case
+                        global_accumulative_recall = global_accumulative_recall + 1.0
+                        global_accumulative_precision = global_accumulative_precision + 1.0
+                        local_accumulative_recall = local_accumulative_recall + 1.0
+                        local_accumulative_precision = local_accumulative_precision + 1.0
+
+                        gt_flag[0, gt_id] = 1
+                        det_flag[0, qualified_tau_candidates] = 1
+                        # recg start
+                        gt_str_cur = global_gt_str[idy][gt_id]
+                        pred_str_cur = global_pred_str[idy][
+                            qualified_tau_candidates[0].tolist()[0]]
+                        if pred_str_cur == gt_str_cur:
+                            hit_str_num += 1
+                        else:
+                            if pred_str_cur.lower() == gt_str_cur.lower():
+                                hit_str_num += 1
+                        # recg end
+                elif (np.sum(local_sigma_table[gt_id, qualified_tau_candidates])
+                      >= tr):
+                    gt_flag[0, gt_id] = 1
+                    det_flag[0, qualified_tau_candidates] = 1
+                    # recg start
+                    gt_str_cur = global_gt_str[idy][gt_id]
+                    pred_str_cur = global_pred_str[idy][
+                        qualified_tau_candidates[0].tolist()[0]]
+                    if pred_str_cur == gt_str_cur:
+                        hit_str_num += 1
+                    else:
+                        if pred_str_cur.lower() == gt_str_cur.lower():
+                            hit_str_num += 1
+                    # recg end
+
+                    global_accumulative_recall = global_accumulative_recall + fsc_k
+                    global_accumulative_precision = global_accumulative_precision + num_qualified_tau_candidates * fsc_k
+
+                    local_accumulative_recall = local_accumulative_recall + fsc_k
+                    local_accumulative_precision = local_accumulative_precision + num_qualified_tau_candidates * fsc_k
+
+        return local_accumulative_recall, local_accumulative_precision, global_accumulative_recall, global_accumulative_precision, gt_flag, det_flag, hit_str_num
+
+    def many_to_one(local_sigma_table, local_tau_table,
+                    local_accumulative_recall, local_accumulative_precision,
+                    global_accumulative_recall, global_accumulative_precision,
+                    gt_flag, det_flag, idy):
+        hit_str_num = 0
+        for det_id in range(num_det):
+            # skip the following if the detection was matched
+            if det_flag[0, det_id] > 0:
+                continue
+
+            non_zero_in_tau = np.where(local_tau_table[:, det_id] > 0)
+            num_non_zero_in_tau = non_zero_in_tau[0].shape[0]
+
+            if num_non_zero_in_tau >= k:
+                ####search for all detections that overlaps with this groundtruth
+                qualified_sigma_candidates = np.where((
+                    local_sigma_table[:, det_id] >= tp) & (gt_flag[0, :] == 0))
+                num_qualified_sigma_candidates = qualified_sigma_candidates[
+                    0].shape[0]
+
+                if num_qualified_sigma_candidates == 1:
+                    if ((local_tau_table[qualified_sigma_candidates, det_id] >=
+                         tp) and
+                        (local_sigma_table[qualified_sigma_candidates, det_id]
+                         >= tr)):
+                        # became an one-to-one case
+                        global_accumulative_recall = global_accumulative_recall + 1.0
+                        global_accumulative_precision = global_accumulative_precision + 1.0
+                        local_accumulative_recall = local_accumulative_recall + 1.0
+                        local_accumulative_precision = local_accumulative_precision + 1.0
+
+                        gt_flag[0, qualified_sigma_candidates] = 1
+                        det_flag[0, det_id] = 1
+                        # recg start
+                        pred_str_cur = global_pred_str[idy][det_id]
+                        gt_len = len(qualified_sigma_candidates[0])
+                        for idx in range(gt_len):
+                            ele_gt_id = qualified_sigma_candidates[0].tolist()[
+                                idx]
+                            if ele_gt_id not in global_gt_str[idy]:
+                                continue
+                            gt_str_cur = global_gt_str[idy][ele_gt_id]
+                            if pred_str_cur == gt_str_cur:
+                                hit_str_num += 1
+                                break
+                            else:
+                                if pred_str_cur.lower() == gt_str_cur.lower():
+                                    hit_str_num += 1
+                                break
+                        # recg end
+                elif (np.sum(local_tau_table[qualified_sigma_candidates,
+                                             det_id]) >= tp):
+                    det_flag[0, det_id] = 1
+                    gt_flag[0, qualified_sigma_candidates] = 1
+                    # recg start
+                    pred_str_cur = global_pred_str[idy][det_id]
+                    gt_len = len(qualified_sigma_candidates[0])
+                    for idx in range(gt_len):
+                        ele_gt_id = qualified_sigma_candidates[0].tolist()[idx]
+                        if ele_gt_id not in global_gt_str[idy]:
+                            continue
+                        gt_str_cur = global_gt_str[idy][ele_gt_id]
+                        if pred_str_cur == gt_str_cur:
+                            hit_str_num += 1
+                            break
+                        else:
+                            if pred_str_cur.lower() == gt_str_cur.lower():
+                                hit_str_num += 1
+                                break
+                    # recg end
+
+                    global_accumulative_recall = global_accumulative_recall + num_qualified_sigma_candidates * fsc_k
+                    global_accumulative_precision = global_accumulative_precision + fsc_k
+
+                    local_accumulative_recall = local_accumulative_recall + num_qualified_sigma_candidates * fsc_k
+                    local_accumulative_precision = local_accumulative_precision + fsc_k
+        return local_accumulative_recall, local_accumulative_precision, global_accumulative_recall, global_accumulative_precision, gt_flag, det_flag, hit_str_num
+
+    for idx in range(len(global_sigma)):
+        local_sigma_table = np.array(global_sigma[idx])
+        local_tau_table = global_tau[idx]
+
+        num_gt = local_sigma_table.shape[0]
+        num_det = local_sigma_table.shape[1]
+
+        total_num_gt = total_num_gt + num_gt
+        total_num_det = total_num_det + num_det
+
+        local_accumulative_recall = 0
+        local_accumulative_precision = 0
+        gt_flag = np.zeros((1, num_gt))
+        det_flag = np.zeros((1, num_det))
+
+        #######first check for one-to-one case##########
+        local_accumulative_recall, local_accumulative_precision, global_accumulative_recall, global_accumulative_precision, \
+        gt_flag, det_flag, hit_str_num = one_to_one(local_sigma_table, local_tau_table,
+                                                    local_accumulative_recall, local_accumulative_precision,
+                                                    global_accumulative_recall, global_accumulative_precision,
+                                                    gt_flag, det_flag, idx)
+
+        hit_str_count += hit_str_num
+        #######then check for one-to-many case##########
+        local_accumulative_recall, local_accumulative_precision, global_accumulative_recall, global_accumulative_precision, \
+        gt_flag, det_flag, hit_str_num = one_to_many(local_sigma_table, local_tau_table,
+                                                     local_accumulative_recall, local_accumulative_precision,
+                                                     global_accumulative_recall, global_accumulative_precision,
+                                                     gt_flag, det_flag, idx)
+        hit_str_count += hit_str_num
+        #######then check for many-to-one case##########
+        local_accumulative_recall, local_accumulative_precision, global_accumulative_recall, global_accumulative_precision, \
+        gt_flag, det_flag, hit_str_num = many_to_one(local_sigma_table, local_tau_table,
+                                                     local_accumulative_recall, local_accumulative_precision,
+                                                     global_accumulative_recall, global_accumulative_precision,
+                                                     gt_flag, det_flag, idx)
+        hit_str_count += hit_str_num
+
+    try:
+        recall = global_accumulative_recall / total_num_gt
+    except ZeroDivisionError:
+        recall = 0
+
+    try:
+        precision = global_accumulative_precision / total_num_det
+    except ZeroDivisionError:
+        precision = 0
+
+    try:
+        f_score = 2 * precision * recall / (precision + recall)
+    except ZeroDivisionError:
+        f_score = 0
+
+    try:
+        seqerr = 1 - float(hit_str_count) / global_accumulative_recall
+    except ZeroDivisionError:
+        seqerr = 1
+
+    try:
+        recall_e2e = float(hit_str_count) / total_num_gt
+    except ZeroDivisionError:
+        recall_e2e = 0
+
+    try:
+        precision_e2e = float(hit_str_count) / total_num_det
+    except ZeroDivisionError:
+        precision_e2e = 0
+
+    try:
+        f_score_e2e = 2 * precision_e2e * recall_e2e / (
+            precision_e2e + recall_e2e)
+    except ZeroDivisionError:
+        f_score_e2e = 0
+
+    final = {
+        'total_num_gt': total_num_gt,
+        'total_num_det': total_num_det,
+        'global_accumulative_recall': global_accumulative_recall,
+        'hit_str_count': hit_str_count,
+        'recall': recall,
+        'precision': precision,
+        'f_score': f_score,
+        'seqerr': seqerr,
+        'recall_e2e': recall_e2e,
+        'precision_e2e': precision_e2e,
+        'f_score_e2e': f_score_e2e
+    }
+    return final

backend/ppocr/utils/e2e_metric/polygon_fast.py

@@ -0,0 +1,83 @@
+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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 numpy as np
+from shapely.geometry import Polygon
+"""
+:param det_x: [1, N] Xs of detection's vertices 
+:param det_y: [1, N] Ys of detection's vertices
+:param gt_x: [1, N] Xs of groundtruth's vertices
+:param gt_y: [1, N] Ys of groundtruth's vertices
+
+##############
+All the calculation of 'AREA' in this script is handled by:
+1) First generating a binary mask with the polygon area filled up with 1's
+2) Summing up all the 1's
+"""
+
+
+def area(x, y):
+    polygon = Polygon(np.stack([x, y], axis=1))
+    return float(polygon.area)
+
+
+def approx_area_of_intersection(det_x, det_y, gt_x, gt_y):
+    """
+    This helper determine if both polygons are intersecting with each others with an approximation method.
+    Area of intersection represented by the minimum bounding rectangular [xmin, ymin, xmax, ymax]
+    """
+    det_ymax = np.max(det_y)
+    det_xmax = np.max(det_x)
+    det_ymin = np.min(det_y)
+    det_xmin = np.min(det_x)
+
+    gt_ymax = np.max(gt_y)
+    gt_xmax = np.max(gt_x)
+    gt_ymin = np.min(gt_y)
+    gt_xmin = np.min(gt_x)
+
+    all_min_ymax = np.minimum(det_ymax, gt_ymax)
+    all_max_ymin = np.maximum(det_ymin, gt_ymin)
+
+    intersect_heights = np.maximum(0.0, (all_min_ymax - all_max_ymin))
+
+    all_min_xmax = np.minimum(det_xmax, gt_xmax)
+    all_max_xmin = np.maximum(det_xmin, gt_xmin)
+    intersect_widths = np.maximum(0.0, (all_min_xmax - all_max_xmin))
+
+    return intersect_heights * intersect_widths
+
+
+def area_of_intersection(det_x, det_y, gt_x, gt_y):
+    p1 = Polygon(np.stack([det_x, det_y], axis=1)).buffer(0)
+    p2 = Polygon(np.stack([gt_x, gt_y], axis=1)).buffer(0)
+    return float(p1.intersection(p2).area)
+
+
+def area_of_union(det_x, det_y, gt_x, gt_y):
+    p1 = Polygon(np.stack([det_x, det_y], axis=1)).buffer(0)
+    p2 = Polygon(np.stack([gt_x, gt_y], axis=1)).buffer(0)
+    return float(p1.union(p2).area)
+
+
+def iou(det_x, det_y, gt_x, gt_y):
+    return area_of_intersection(det_x, det_y, gt_x, gt_y) / (
+        area_of_union(det_x, det_y, gt_x, gt_y) + 1.0)
+
+
+def iod(det_x, det_y, gt_x, gt_y):
+    """
+    This helper determine the fraction of intersection area over detection area
+    """
+    return area_of_intersection(det_x, det_y, gt_x, gt_y) / (
+        area(det_x, det_y) + 1.0)

backend/ppocr/utils/e2e_utils/extract_batchsize.py

@@ -0,0 +1,87 @@
+import paddle
+import numpy as np
+import copy
+
+
+def org_tcl_rois(batch_size, pos_lists, pos_masks, label_lists, tcl_bs):
+    """
+    """
+    pos_lists_, pos_masks_, label_lists_ = [], [], []
+    img_bs = batch_size
+    ngpu = int(batch_size / img_bs)
+    img_ids = np.array(pos_lists, dtype=np.int32)[:, 0, 0].copy()
+    pos_lists_split, pos_masks_split, label_lists_split = [], [], []
+    for i in range(ngpu):
+        pos_lists_split.append([])
+        pos_masks_split.append([])
+        label_lists_split.append([])
+
+    for i in range(img_ids.shape[0]):
+        img_id = img_ids[i]
+        gpu_id = int(img_id / img_bs)
+        img_id = img_id % img_bs
+        pos_list = pos_lists[i].copy()
+        pos_list[:, 0] = img_id
+        pos_lists_split[gpu_id].append(pos_list)
+        pos_masks_split[gpu_id].append(pos_masks[i].copy())
+        label_lists_split[gpu_id].append(copy.deepcopy(label_lists[i]))
+    # repeat or delete
+    for i in range(ngpu):
+        vp_len = len(pos_lists_split[i])
+        if vp_len <= tcl_bs:
+            for j in range(0, tcl_bs - vp_len):
+                pos_list = pos_lists_split[i][j].copy()
+                pos_lists_split[i].append(pos_list)
+                pos_mask = pos_masks_split[i][j].copy()
+                pos_masks_split[i].append(pos_mask)
+                label_list = copy.deepcopy(label_lists_split[i][j])
+                label_lists_split[i].append(label_list)
+        else:
+            for j in range(0, vp_len - tcl_bs):
+                c_len = len(pos_lists_split[i])
+                pop_id = np.random.permutation(c_len)[0]
+                pos_lists_split[i].pop(pop_id)
+                pos_masks_split[i].pop(pop_id)
+                label_lists_split[i].pop(pop_id)
+    # merge
+    for i in range(ngpu):
+        pos_lists_.extend(pos_lists_split[i])
+        pos_masks_.extend(pos_masks_split[i])
+        label_lists_.extend(label_lists_split[i])
+    return pos_lists_, pos_masks_, label_lists_
+
+
+def pre_process(label_list, pos_list, pos_mask, max_text_length, max_text_nums,
+                pad_num, tcl_bs):
+    label_list = label_list.numpy()
+    batch, _, _, _ = label_list.shape
+    pos_list = pos_list.numpy()
+    pos_mask = pos_mask.numpy()
+    pos_list_t = []
+    pos_mask_t = []
+    label_list_t = []
+    for i in range(batch):
+        for j in range(max_text_nums):
+            if pos_mask[i, j].any():
+                pos_list_t.append(pos_list[i][j])
+                pos_mask_t.append(pos_mask[i][j])
+                label_list_t.append(label_list[i][j])
+    pos_list, pos_mask, label_list = org_tcl_rois(batch, pos_list_t, pos_mask_t,
+                                                  label_list_t, tcl_bs)
+    label = []
+    tt = [l.tolist() for l in label_list]
+    for i in range(tcl_bs):
+        k = 0
+        for j in range(max_text_length):
+            if tt[i][j][0] != pad_num:
+                k += 1
+            else:
+                break
+        label.append(k)
+    label = paddle.to_tensor(label)
+    label = paddle.cast(label, dtype='int64')
+    pos_list = paddle.to_tensor(pos_list)
+    pos_mask = paddle.to_tensor(pos_mask)
+    label_list = paddle.squeeze(paddle.to_tensor(label_list), axis=2)
+    label_list = paddle.cast(label_list, dtype='int32')
+    return pos_list, pos_mask, label_list, label

backend/ppocr/utils/e2e_utils/extract_textpoint_fast.py

@@ -0,0 +1,457 @@
+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+"""Contains various CTC decoders."""
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import cv2
+import math
+
+import numpy as np
+from itertools import groupby
+from skimage.morphology._skeletonize import thin
+
+
+def get_dict(character_dict_path):
+    character_str = ""
+    with open(character_dict_path, "rb") as fin:
+        lines = fin.readlines()
+        for line in lines:
+            line = line.decode('utf-8').strip("\n").strip("\r\n")
+            character_str += line
+        dict_character = list(character_str)
+    return dict_character
+
+
+def softmax(logits):
+    """
+    logits: N x d
+    """
+    max_value = np.max(logits, axis=1, keepdims=True)
+    exp = np.exp(logits - max_value)
+    exp_sum = np.sum(exp, axis=1, keepdims=True)
+    dist = exp / exp_sum
+    return dist
+
+
+def get_keep_pos_idxs(labels, remove_blank=None):
+    """
+    Remove duplicate and get pos idxs of keep items.
+    The value of keep_blank should be [None, 95].
+    """
+    duplicate_len_list = []
+    keep_pos_idx_list = []
+    keep_char_idx_list = []
+    for k, v_ in groupby(labels):
+        current_len = len(list(v_))
+        if k != remove_blank:
+            current_idx = int(sum(duplicate_len_list) + current_len // 2)
+            keep_pos_idx_list.append(current_idx)
+            keep_char_idx_list.append(k)
+        duplicate_len_list.append(current_len)
+    return keep_char_idx_list, keep_pos_idx_list
+
+
+def remove_blank(labels, blank=0):
+    new_labels = [x for x in labels if x != blank]
+    return new_labels
+
+
+def insert_blank(labels, blank=0):
+    new_labels = [blank]
+    for l in labels:
+        new_labels += [l, blank]
+    return new_labels
+
+
+def ctc_greedy_decoder(probs_seq, blank=95, keep_blank_in_idxs=True):
+    """
+    CTC greedy (best path) decoder.
+    """
+    raw_str = np.argmax(np.array(probs_seq), axis=1)
+    remove_blank_in_pos = None if keep_blank_in_idxs else blank
+    dedup_str, keep_idx_list = get_keep_pos_idxs(
+        raw_str, remove_blank=remove_blank_in_pos)
+    dst_str = remove_blank(dedup_str, blank=blank)
+    return dst_str, keep_idx_list
+
+
+def instance_ctc_greedy_decoder(gather_info, logits_map, pts_num=4):
+    _, _, C = logits_map.shape
+    ys, xs = zip(*gather_info)
+    logits_seq = logits_map[list(ys), list(xs)]
+    probs_seq = logits_seq
+    labels = np.argmax(probs_seq, axis=1)
+    dst_str = [k for k, v_ in groupby(labels) if k != C - 1]
+    detal = len(gather_info) // (pts_num - 1)
+    keep_idx_list = [0] + [detal * (i + 1) for i in range(pts_num - 2)] + [-1]
+    keep_gather_list = [gather_info[idx] for idx in keep_idx_list]
+    return dst_str, keep_gather_list
+
+
+def ctc_decoder_for_image(gather_info_list,
+                          logits_map,
+                          Lexicon_Table,
+                          pts_num=6):
+    """
+    CTC decoder using multiple processes.
+    """
+    decoder_str = []
+    decoder_xys = []
+    for gather_info in gather_info_list:
+        if len(gather_info) < pts_num:
+            continue
+        dst_str, xys_list = instance_ctc_greedy_decoder(
+            gather_info, logits_map, pts_num=pts_num)
+        dst_str_readable = ''.join([Lexicon_Table[idx] for idx in dst_str])
+        if len(dst_str_readable) < 2:
+            continue
+        decoder_str.append(dst_str_readable)
+        decoder_xys.append(xys_list)
+    return decoder_str, decoder_xys
+
+
+def sort_with_direction(pos_list, f_direction):
+    """
+    f_direction: h x w x 2
+    pos_list: [[y, x], [y, x], [y, x] ...]
+    """
+
+    def sort_part_with_direction(pos_list, point_direction):
+        pos_list = np.array(pos_list).reshape(-1, 2)
+        point_direction = np.array(point_direction).reshape(-1, 2)
+        average_direction = np.mean(point_direction, axis=0, keepdims=True)
+        pos_proj_leng = np.sum(pos_list * average_direction, axis=1)
+        sorted_list = pos_list[np.argsort(pos_proj_leng)].tolist()
+        sorted_direction = point_direction[np.argsort(pos_proj_leng)].tolist()
+        return sorted_list, sorted_direction
+
+    pos_list = np.array(pos_list).reshape(-1, 2)
+    point_direction = f_direction[pos_list[:, 0], pos_list[:, 1]]  # x, y
+    point_direction = point_direction[:, ::-1]  # x, y -> y, x
+    sorted_point, sorted_direction = sort_part_with_direction(pos_list,
+                                                              point_direction)
+
+    point_num = len(sorted_point)
+    if point_num >= 16:
+        middle_num = point_num // 2
+        first_part_point = sorted_point[:middle_num]
+        first_point_direction = sorted_direction[:middle_num]
+        sorted_fist_part_point, sorted_fist_part_direction = sort_part_with_direction(
+            first_part_point, first_point_direction)
+
+        last_part_point = sorted_point[middle_num:]
+        last_point_direction = sorted_direction[middle_num:]
+        sorted_last_part_point, sorted_last_part_direction = sort_part_with_direction(
+            last_part_point, last_point_direction)
+        sorted_point = sorted_fist_part_point + sorted_last_part_point
+        sorted_direction = sorted_fist_part_direction + sorted_last_part_direction
+
+    return sorted_point, np.array(sorted_direction)
+
+
+def add_id(pos_list, image_id=0):
+    """
+    Add id for gather feature, for inference.
+    """
+    new_list = []
+    for item in pos_list:
+        new_list.append((image_id, item[0], item[1]))
+    return new_list
+
+
+def sort_and_expand_with_direction(pos_list, f_direction):
+    """
+    f_direction: h x w x 2
+    pos_list: [[y, x], [y, x], [y, x] ...]
+    """
+    h, w, _ = f_direction.shape
+    sorted_list, point_direction = sort_with_direction(pos_list, f_direction)
+
+    point_num = len(sorted_list)
+    sub_direction_len = max(point_num // 3, 2)
+    left_direction = point_direction[:sub_direction_len, :]
+    right_dirction = point_direction[point_num - sub_direction_len:, :]
+
+    left_average_direction = -np.mean(left_direction, axis=0, keepdims=True)
+    left_average_len = np.linalg.norm(left_average_direction)
+    left_start = np.array(sorted_list[0])
+    left_step = left_average_direction / (left_average_len + 1e-6)
+
+    right_average_direction = np.mean(right_dirction, axis=0, keepdims=True)
+    right_average_len = np.linalg.norm(right_average_direction)
+    right_step = right_average_direction / (right_average_len + 1e-6)
+    right_start = np.array(sorted_list[-1])
+
+    append_num = max(
+        int((left_average_len + right_average_len) / 2.0 * 0.15), 1)
+    left_list = []
+    right_list = []
+    for i in range(append_num):
+        ly, lx = np.round(left_start + left_step * (i + 1)).flatten().astype(
+            'int32').tolist()
+        if ly < h and lx < w and (ly, lx) not in left_list:
+            left_list.append((ly, lx))
+        ry, rx = np.round(right_start + right_step * (i + 1)).flatten().astype(
+            'int32').tolist()
+        if ry < h and rx < w and (ry, rx) not in right_list:
+            right_list.append((ry, rx))
+
+    all_list = left_list[::-1] + sorted_list + right_list
+    return all_list
+
+
+def sort_and_expand_with_direction_v2(pos_list, f_direction, binary_tcl_map):
+    """
+    f_direction: h x w x 2
+    pos_list: [[y, x], [y, x], [y, x] ...]
+    binary_tcl_map: h x w
+    """
+    h, w, _ = f_direction.shape
+    sorted_list, point_direction = sort_with_direction(pos_list, f_direction)
+
+    point_num = len(sorted_list)
+    sub_direction_len = max(point_num // 3, 2)
+    left_direction = point_direction[:sub_direction_len, :]
+    right_dirction = point_direction[point_num - sub_direction_len:, :]
+
+    left_average_direction = -np.mean(left_direction, axis=0, keepdims=True)
+    left_average_len = np.linalg.norm(left_average_direction)
+    left_start = np.array(sorted_list[0])
+    left_step = left_average_direction / (left_average_len + 1e-6)
+
+    right_average_direction = np.mean(right_dirction, axis=0, keepdims=True)
+    right_average_len = np.linalg.norm(right_average_direction)
+    right_step = right_average_direction / (right_average_len + 1e-6)
+    right_start = np.array(sorted_list[-1])
+
+    append_num = max(
+        int((left_average_len + right_average_len) / 2.0 * 0.15), 1)
+    max_append_num = 2 * append_num
+
+    left_list = []
+    right_list = []
+    for i in range(max_append_num):
+        ly, lx = np.round(left_start + left_step * (i + 1)).flatten().astype(
+            'int32').tolist()
+        if ly < h and lx < w and (ly, lx) not in left_list:
+            if binary_tcl_map[ly, lx] > 0.5:
+                left_list.append((ly, lx))
+            else:
+                break
+
+    for i in range(max_append_num):
+        ry, rx = np.round(right_start + right_step * (i + 1)).flatten().astype(
+            'int32').tolist()
+        if ry < h and rx < w and (ry, rx) not in right_list:
+            if binary_tcl_map[ry, rx] > 0.5:
+                right_list.append((ry, rx))
+            else:
+                break
+
+    all_list = left_list[::-1] + sorted_list + right_list
+    return all_list
+
+
+def point_pair2poly(point_pair_list):
+    """
+    Transfer vertical point_pairs into poly point in clockwise.
+    """
+    point_num = len(point_pair_list) * 2
+    point_list = [0] * point_num
+    for idx, point_pair in enumerate(point_pair_list):
+        point_list[idx] = point_pair[0]
+        point_list[point_num - 1 - idx] = point_pair[1]
+    return np.array(point_list).reshape(-1, 2)
+
+
+def shrink_quad_along_width(quad, begin_width_ratio=0., end_width_ratio=1.):
+    ratio_pair = np.array(
+        [[begin_width_ratio], [end_width_ratio]], dtype=np.float32)
+    p0_1 = quad[0] + (quad[1] - quad[0]) * ratio_pair
+    p3_2 = quad[3] + (quad[2] - quad[3]) * ratio_pair
+    return np.array([p0_1[0], p0_1[1], p3_2[1], p3_2[0]])
+
+
+def expand_poly_along_width(poly, shrink_ratio_of_width=0.3):
+    """
+    expand poly along width.
+    """
+    point_num = poly.shape[0]
+    left_quad = np.array(
+        [poly[0], poly[1], poly[-2], poly[-1]], dtype=np.float32)
+    left_ratio = -shrink_ratio_of_width * np.linalg.norm(left_quad[0] - left_quad[3]) / \
+                 (np.linalg.norm(left_quad[0] - left_quad[1]) + 1e-6)
+    left_quad_expand = shrink_quad_along_width(left_quad, left_ratio, 1.0)
+    right_quad = np.array(
+        [
+            poly[point_num // 2 - 2], poly[point_num // 2 - 1],
+            poly[point_num // 2], poly[point_num // 2 + 1]
+        ],
+        dtype=np.float32)
+    right_ratio = 1.0 + shrink_ratio_of_width * np.linalg.norm(right_quad[0] - right_quad[3]) / \
+                  (np.linalg.norm(right_quad[0] - right_quad[1]) + 1e-6)
+    right_quad_expand = shrink_quad_along_width(right_quad, 0.0, right_ratio)
+    poly[0] = left_quad_expand[0]
+    poly[-1] = left_quad_expand[-1]
+    poly[point_num // 2 - 1] = right_quad_expand[1]
+    poly[point_num // 2] = right_quad_expand[2]
+    return poly
+
+
+def restore_poly(instance_yxs_list, seq_strs, p_border, ratio_w, ratio_h, src_w,
+                 src_h, valid_set):
+    poly_list = []
+    keep_str_list = []
+    for yx_center_line, keep_str in zip(instance_yxs_list, seq_strs):
+        if len(keep_str) < 2:
+            print('--> too short, {}'.format(keep_str))
+            continue
+
+        offset_expand = 1.0
+        if valid_set == 'totaltext':
+            offset_expand = 1.2
+
+        point_pair_list = []
+        for y, x in yx_center_line:
+            offset = p_border[:, y, x].reshape(2, 2) * offset_expand
+            ori_yx = np.array([y, x], dtype=np.float32)
+            point_pair = (ori_yx + offset)[:, ::-1] * 4.0 / np.array(
+                [ratio_w, ratio_h]).reshape(-1, 2)
+            point_pair_list.append(point_pair)
+
+        detected_poly = point_pair2poly(point_pair_list)
+        detected_poly = expand_poly_along_width(
+            detected_poly, shrink_ratio_of_width=0.2)
+        detected_poly[:, 0] = np.clip(detected_poly[:, 0], a_min=0, a_max=src_w)
+        detected_poly[:, 1] = np.clip(detected_poly[:, 1], a_min=0, a_max=src_h)
+
+        keep_str_list.append(keep_str)
+        if valid_set == 'partvgg':
+            middle_point = len(detected_poly) // 2
+            detected_poly = detected_poly[
+                [0, middle_point - 1, middle_point, -1], :]
+            poly_list.append(detected_poly)
+        elif valid_set == 'totaltext':
+            poly_list.append(detected_poly)
+        else:
+            print('--> Not supported format.')
+            exit(-1)
+    return poly_list, keep_str_list
+
+
+def generate_pivot_list_fast(p_score,
+                             p_char_maps,
+                             f_direction,
+                             Lexicon_Table,
+                             score_thresh=0.5):
+    """
+    return center point and end point of TCL instance; filter with the char maps;
+    """
+    p_score = p_score[0]
+    f_direction = f_direction.transpose(1, 2, 0)
+    p_tcl_map = (p_score > score_thresh) * 1.0
+    skeleton_map = thin(p_tcl_map.astype(np.uint8))
+    instance_count, instance_label_map = cv2.connectedComponents(
+        skeleton_map.astype(np.uint8), connectivity=8)
+
+    # get TCL Instance
+    all_pos_yxs = []
+    if instance_count > 0:
+        for instance_id in range(1, instance_count):
+            pos_list = []
+            ys, xs = np.where(instance_label_map == instance_id)
+            pos_list = list(zip(ys, xs))
+
+            if len(pos_list) < 3:
+                continue
+
+            pos_list_sorted = sort_and_expand_with_direction_v2(
+                pos_list, f_direction, p_tcl_map)
+            all_pos_yxs.append(pos_list_sorted)
+
+    p_char_maps = p_char_maps.transpose([1, 2, 0])
+    decoded_str, keep_yxs_list = ctc_decoder_for_image(
+        all_pos_yxs, logits_map=p_char_maps, Lexicon_Table=Lexicon_Table)
+    return keep_yxs_list, decoded_str
+
+
+def extract_main_direction(pos_list, f_direction):
+    """
+    f_direction: h x w x 2
+    pos_list: [[y, x], [y, x], [y, x] ...]
+    """
+    pos_list = np.array(pos_list)
+    point_direction = f_direction[pos_list[:, 0], pos_list[:, 1]]
+    point_direction = point_direction[:, ::-1]  # x, y -> y, x
+    average_direction = np.mean(point_direction, axis=0, keepdims=True)
+    average_direction = average_direction / (
+        np.linalg.norm(average_direction) + 1e-6)
+    return average_direction
+
+
+def sort_by_direction_with_image_id_deprecated(pos_list, f_direction):
+    """
+    f_direction: h x w x 2
+    pos_list: [[id, y, x], [id, y, x], [id, y, x] ...]
+    """
+    pos_list_full = np.array(pos_list).reshape(-1, 3)
+    pos_list = pos_list_full[:, 1:]
+    point_direction = f_direction[pos_list[:, 0], pos_list[:, 1]]  # x, y
+    point_direction = point_direction[:, ::-1]  # x, y -> y, x
+    average_direction = np.mean(point_direction, axis=0, keepdims=True)
+    pos_proj_leng = np.sum(pos_list * average_direction, axis=1)
+    sorted_list = pos_list_full[np.argsort(pos_proj_leng)].tolist()
+    return sorted_list
+
+
+def sort_by_direction_with_image_id(pos_list, f_direction):
+    """
+    f_direction: h x w x 2
+    pos_list: [[y, x], [y, x], [y, x] ...]
+    """
+
+    def sort_part_with_direction(pos_list_full, point_direction):
+        pos_list_full = np.array(pos_list_full).reshape(-1, 3)
+        pos_list = pos_list_full[:, 1:]
+        point_direction = np.array(point_direction).reshape(-1, 2)
+        average_direction = np.mean(point_direction, axis=0, keepdims=True)
+        pos_proj_leng = np.sum(pos_list * average_direction, axis=1)
+        sorted_list = pos_list_full[np.argsort(pos_proj_leng)].tolist()
+        sorted_direction = point_direction[np.argsort(pos_proj_leng)].tolist()
+        return sorted_list, sorted_direction
+
+    pos_list = np.array(pos_list).reshape(-1, 3)
+    point_direction = f_direction[pos_list[:, 1], pos_list[:, 2]]  # x, y
+    point_direction = point_direction[:, ::-1]  # x, y -> y, x
+    sorted_point, sorted_direction = sort_part_with_direction(pos_list,
+                                                              point_direction)
+
+    point_num = len(sorted_point)
+    if point_num >= 16:
+        middle_num = point_num // 2
+        first_part_point = sorted_point[:middle_num]
+        first_point_direction = sorted_direction[:middle_num]
+        sorted_fist_part_point, sorted_fist_part_direction = sort_part_with_direction(
+            first_part_point, first_point_direction)
+
+        last_part_point = sorted_point[middle_num:]
+        last_point_direction = sorted_direction[middle_num:]
+        sorted_last_part_point, sorted_last_part_direction = sort_part_with_direction(
+            last_part_point, last_point_direction)
+        sorted_point = sorted_fist_part_point + sorted_last_part_point
+        sorted_direction = sorted_fist_part_direction + sorted_last_part_direction
+
+    return sorted_point

backend/ppocr/utils/e2e_utils/extract_textpoint_slow.py

@@ -0,0 +1,592 @@
+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+"""Contains various CTC decoders."""
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import cv2
+import math
+
+import numpy as np
+from itertools import groupby
+from skimage.morphology._skeletonize import thin
+
+
+def get_dict(character_dict_path):
+    character_str = ""
+    with open(character_dict_path, "rb") as fin:
+        lines = fin.readlines()
+        for line in lines:
+            line = line.decode('utf-8').strip("\n").strip("\r\n")
+            character_str += line
+        dict_character = list(character_str)
+    return dict_character
+
+
+def point_pair2poly(point_pair_list):
+    """
+    Transfer vertical point_pairs into poly point in clockwise.
+    """
+    pair_length_list = []
+    for point_pair in point_pair_list:
+        pair_length = np.linalg.norm(point_pair[0] - point_pair[1])
+        pair_length_list.append(pair_length)
+    pair_length_list = np.array(pair_length_list)
+    pair_info = (pair_length_list.max(), pair_length_list.min(),
+                 pair_length_list.mean())
+
+    point_num = len(point_pair_list) * 2
+    point_list = [0] * point_num
+    for idx, point_pair in enumerate(point_pair_list):
+        point_list[idx] = point_pair[0]
+        point_list[point_num - 1 - idx] = point_pair[1]
+    return np.array(point_list).reshape(-1, 2), pair_info
+
+
+def shrink_quad_along_width(quad, begin_width_ratio=0., end_width_ratio=1.):
+    """
+    Generate shrink_quad_along_width.
+    """
+    ratio_pair = np.array(
+        [[begin_width_ratio], [end_width_ratio]], dtype=np.float32)
+    p0_1 = quad[0] + (quad[1] - quad[0]) * ratio_pair
+    p3_2 = quad[3] + (quad[2] - quad[3]) * ratio_pair
+    return np.array([p0_1[0], p0_1[1], p3_2[1], p3_2[0]])
+
+
+def expand_poly_along_width(poly, shrink_ratio_of_width=0.3):
+    """
+    expand poly along width.
+    """
+    point_num = poly.shape[0]
+    left_quad = np.array(
+        [poly[0], poly[1], poly[-2], poly[-1]], dtype=np.float32)
+    left_ratio = -shrink_ratio_of_width * np.linalg.norm(left_quad[0] - left_quad[3]) / \
+                 (np.linalg.norm(left_quad[0] - left_quad[1]) + 1e-6)
+    left_quad_expand = shrink_quad_along_width(left_quad, left_ratio, 1.0)
+    right_quad = np.array(
+        [
+            poly[point_num // 2 - 2], poly[point_num // 2 - 1],
+            poly[point_num // 2], poly[point_num // 2 + 1]
+        ],
+        dtype=np.float32)
+    right_ratio = 1.0 + \
+                  shrink_ratio_of_width * np.linalg.norm(right_quad[0] - right_quad[3]) / \
+                  (np.linalg.norm(right_quad[0] - right_quad[1]) + 1e-6)
+    right_quad_expand = shrink_quad_along_width(right_quad, 0.0, right_ratio)
+    poly[0] = left_quad_expand[0]
+    poly[-1] = left_quad_expand[-1]
+    poly[point_num // 2 - 1] = right_quad_expand[1]
+    poly[point_num // 2] = right_quad_expand[2]
+    return poly
+
+
+def softmax(logits):
+    """
+    logits: N x d
+    """
+    max_value = np.max(logits, axis=1, keepdims=True)
+    exp = np.exp(logits - max_value)
+    exp_sum = np.sum(exp, axis=1, keepdims=True)
+    dist = exp / exp_sum
+    return dist
+
+
+def get_keep_pos_idxs(labels, remove_blank=None):
+    """
+    Remove duplicate and get pos idxs of keep items.
+    The value of keep_blank should be [None, 95].
+    """
+    duplicate_len_list = []
+    keep_pos_idx_list = []
+    keep_char_idx_list = []
+    for k, v_ in groupby(labels):
+        current_len = len(list(v_))
+        if k != remove_blank:
+            current_idx = int(sum(duplicate_len_list) + current_len // 2)
+            keep_pos_idx_list.append(current_idx)
+            keep_char_idx_list.append(k)
+        duplicate_len_list.append(current_len)
+    return keep_char_idx_list, keep_pos_idx_list
+
+
+def remove_blank(labels, blank=0):
+    new_labels = [x for x in labels if x != blank]
+    return new_labels
+
+
+def insert_blank(labels, blank=0):
+    new_labels = [blank]
+    for l in labels:
+        new_labels += [l, blank]
+    return new_labels
+
+
+def ctc_greedy_decoder(probs_seq, blank=95, keep_blank_in_idxs=True):
+    """
+    CTC greedy (best path) decoder.
+    """
+    raw_str = np.argmax(np.array(probs_seq), axis=1)
+    remove_blank_in_pos = None if keep_blank_in_idxs else blank
+    dedup_str, keep_idx_list = get_keep_pos_idxs(
+        raw_str, remove_blank=remove_blank_in_pos)
+    dst_str = remove_blank(dedup_str, blank=blank)
+    return dst_str, keep_idx_list
+
+
+def instance_ctc_greedy_decoder(gather_info,
+                                logits_map,
+                                keep_blank_in_idxs=True):
+    """
+    gather_info: [[x, y], [x, y] ...]
+    logits_map: H x W X (n_chars + 1)
+    """
+    _, _, C = logits_map.shape
+    ys, xs = zip(*gather_info)
+    logits_seq = logits_map[list(ys), list(xs)]  # n x 96
+    probs_seq = softmax(logits_seq)
+    dst_str, keep_idx_list = ctc_greedy_decoder(
+        probs_seq, blank=C - 1, keep_blank_in_idxs=keep_blank_in_idxs)
+    keep_gather_list = [gather_info[idx] for idx in keep_idx_list]
+    return dst_str, keep_gather_list
+
+
+def ctc_decoder_for_image(gather_info_list, logits_map,
+                          keep_blank_in_idxs=True):
+    """
+    CTC decoder using multiple processes.
+    """
+    decoder_results = []
+    for gather_info in gather_info_list:
+        res = instance_ctc_greedy_decoder(
+            gather_info, logits_map, keep_blank_in_idxs=keep_blank_in_idxs)
+        decoder_results.append(res)
+    return decoder_results
+
+
+def sort_with_direction(pos_list, f_direction):
+    """
+    f_direction: h x w x 2
+    pos_list: [[y, x], [y, x], [y, x] ...]
+    """
+
+    def sort_part_with_direction(pos_list, point_direction):
+        pos_list = np.array(pos_list).reshape(-1, 2)
+        point_direction = np.array(point_direction).reshape(-1, 2)
+        average_direction = np.mean(point_direction, axis=0, keepdims=True)
+        pos_proj_leng = np.sum(pos_list * average_direction, axis=1)
+        sorted_list = pos_list[np.argsort(pos_proj_leng)].tolist()
+        sorted_direction = point_direction[np.argsort(pos_proj_leng)].tolist()
+        return sorted_list, sorted_direction
+
+    pos_list = np.array(pos_list).reshape(-1, 2)
+    point_direction = f_direction[pos_list[:, 0], pos_list[:, 1]]  # x, y
+    point_direction = point_direction[:, ::-1]  # x, y -> y, x
+    sorted_point, sorted_direction = sort_part_with_direction(pos_list,
+                                                              point_direction)
+
+    point_num = len(sorted_point)
+    if point_num >= 16:
+        middle_num = point_num // 2
+        first_part_point = sorted_point[:middle_num]
+        first_point_direction = sorted_direction[:middle_num]
+        sorted_fist_part_point, sorted_fist_part_direction = sort_part_with_direction(
+            first_part_point, first_point_direction)
+
+        last_part_point = sorted_point[middle_num:]
+        last_point_direction = sorted_direction[middle_num:]
+        sorted_last_part_point, sorted_last_part_direction = sort_part_with_direction(
+            last_part_point, last_point_direction)
+        sorted_point = sorted_fist_part_point + sorted_last_part_point
+        sorted_direction = sorted_fist_part_direction + sorted_last_part_direction
+
+    return sorted_point, np.array(sorted_direction)
+
+
+def add_id(pos_list, image_id=0):
+    """
+    Add id for gather feature, for inference.
+    """
+    new_list = []
+    for item in pos_list:
+        new_list.append((image_id, item[0], item[1]))
+    return new_list
+
+
+def sort_and_expand_with_direction(pos_list, f_direction):
+    """
+    f_direction: h x w x 2
+    pos_list: [[y, x], [y, x], [y, x] ...]
+    """
+    h, w, _ = f_direction.shape
+    sorted_list, point_direction = sort_with_direction(pos_list, f_direction)
+
+    # expand along
+    point_num = len(sorted_list)
+    sub_direction_len = max(point_num // 3, 2)
+    left_direction = point_direction[:sub_direction_len, :]
+    right_dirction = point_direction[point_num - sub_direction_len:, :]
+
+    left_average_direction = -np.mean(left_direction, axis=0, keepdims=True)
+    left_average_len = np.linalg.norm(left_average_direction)
+    left_start = np.array(sorted_list[0])
+    left_step = left_average_direction / (left_average_len + 1e-6)
+
+    right_average_direction = np.mean(right_dirction, axis=0, keepdims=True)
+    right_average_len = np.linalg.norm(right_average_direction)
+    right_step = right_average_direction / (right_average_len + 1e-6)
+    right_start = np.array(sorted_list[-1])
+
+    append_num = max(
+        int((left_average_len + right_average_len) / 2.0 * 0.15), 1)
+    left_list = []
+    right_list = []
+    for i in range(append_num):
+        ly, lx = np.round(left_start + left_step * (i + 1)).flatten().astype(
+            'int32').tolist()
+        if ly < h and lx < w and (ly, lx) not in left_list:
+            left_list.append((ly, lx))
+        ry, rx = np.round(right_start + right_step * (i + 1)).flatten().astype(
+            'int32').tolist()
+        if ry < h and rx < w and (ry, rx) not in right_list:
+            right_list.append((ry, rx))
+
+    all_list = left_list[::-1] + sorted_list + right_list
+    return all_list
+
+
+def sort_and_expand_with_direction_v2(pos_list, f_direction, binary_tcl_map):
+    """
+    f_direction: h x w x 2
+    pos_list: [[y, x], [y, x], [y, x] ...]
+    binary_tcl_map: h x w
+    """
+    h, w, _ = f_direction.shape
+    sorted_list, point_direction = sort_with_direction(pos_list, f_direction)
+
+    # expand along
+    point_num = len(sorted_list)
+    sub_direction_len = max(point_num // 3, 2)
+    left_direction = point_direction[:sub_direction_len, :]
+    right_dirction = point_direction[point_num - sub_direction_len:, :]
+
+    left_average_direction = -np.mean(left_direction, axis=0, keepdims=True)
+    left_average_len = np.linalg.norm(left_average_direction)
+    left_start = np.array(sorted_list[0])
+    left_step = left_average_direction / (left_average_len + 1e-6)
+
+    right_average_direction = np.mean(right_dirction, axis=0, keepdims=True)
+    right_average_len = np.linalg.norm(right_average_direction)
+    right_step = right_average_direction / (right_average_len + 1e-6)
+    right_start = np.array(sorted_list[-1])
+
+    append_num = max(
+        int((left_average_len + right_average_len) / 2.0 * 0.15), 1)
+    max_append_num = 2 * append_num
+
+    left_list = []
+    right_list = []
+    for i in range(max_append_num):
+        ly, lx = np.round(left_start + left_step * (i + 1)).flatten().astype(
+            'int32').tolist()
+        if ly < h and lx < w and (ly, lx) not in left_list:
+            if binary_tcl_map[ly, lx] > 0.5:
+                left_list.append((ly, lx))
+            else:
+                break
+
+    for i in range(max_append_num):
+        ry, rx = np.round(right_start + right_step * (i + 1)).flatten().astype(
+            'int32').tolist()
+        if ry < h and rx < w and (ry, rx) not in right_list:
+            if binary_tcl_map[ry, rx] > 0.5:
+                right_list.append((ry, rx))
+            else:
+                break
+
+    all_list = left_list[::-1] + sorted_list + right_list
+    return all_list
+
+
+def generate_pivot_list_curved(p_score,
+                               p_char_maps,
+                               f_direction,
+                               score_thresh=0.5,
+                               is_expand=True,
+                               is_backbone=False,
+                               image_id=0):
+    """
+    return center point and end point of TCL instance; filter with the char maps;
+    """
+    p_score = p_score[0]
+    f_direction = f_direction.transpose(1, 2, 0)
+    p_tcl_map = (p_score > score_thresh) * 1.0
+    skeleton_map = thin(p_tcl_map)
+    instance_count, instance_label_map = cv2.connectedComponents(
+        skeleton_map.astype(np.uint8), connectivity=8)
+
+    # get TCL Instance
+    all_pos_yxs = []
+    center_pos_yxs = []
+    end_points_yxs = []
+    instance_center_pos_yxs = []
+    pred_strs = []
+    if instance_count > 0:
+        for instance_id in range(1, instance_count):
+            pos_list = []
+            ys, xs = np.where(instance_label_map == instance_id)
+            pos_list = list(zip(ys, xs))
+
+            ### FIX-ME, eliminate outlier
+            if len(pos_list) < 3:
+                continue
+
+            if is_expand:
+                pos_list_sorted = sort_and_expand_with_direction_v2(
+                    pos_list, f_direction, p_tcl_map)
+            else:
+                pos_list_sorted, _ = sort_with_direction(pos_list, f_direction)
+            all_pos_yxs.append(pos_list_sorted)
+
+    # use decoder to filter backgroud points.
+    p_char_maps = p_char_maps.transpose([1, 2, 0])
+    decode_res = ctc_decoder_for_image(
+        all_pos_yxs, logits_map=p_char_maps, keep_blank_in_idxs=True)
+    for decoded_str, keep_yxs_list in decode_res:
+        if is_backbone:
+            keep_yxs_list_with_id = add_id(keep_yxs_list, image_id=image_id)
+            instance_center_pos_yxs.append(keep_yxs_list_with_id)
+            pred_strs.append(decoded_str)
+        else:
+            end_points_yxs.extend((keep_yxs_list[0], keep_yxs_list[-1]))
+            center_pos_yxs.extend(keep_yxs_list)
+
+    if is_backbone:
+        return pred_strs, instance_center_pos_yxs
+    else:
+        return center_pos_yxs, end_points_yxs
+
+
+def generate_pivot_list_horizontal(p_score,
+                                   p_char_maps,
+                                   f_direction,
+                                   score_thresh=0.5,
+                                   is_backbone=False,
+                                   image_id=0):
+    """
+    return center point and end point of TCL instance; filter with the char maps;
+    """
+    p_score = p_score[0]
+    f_direction = f_direction.transpose(1, 2, 0)
+    p_tcl_map_bi = (p_score > score_thresh) * 1.0
+    instance_count, instance_label_map = cv2.connectedComponents(
+        p_tcl_map_bi.astype(np.uint8), connectivity=8)
+
+    # get TCL Instance
+    all_pos_yxs = []
+    center_pos_yxs = []
+    end_points_yxs = []
+    instance_center_pos_yxs = []
+
+    if instance_count > 0:
+        for instance_id in range(1, instance_count):
+            pos_list = []
+            ys, xs = np.where(instance_label_map == instance_id)
+            pos_list = list(zip(ys, xs))
+
+            ### FIX-ME, eliminate outlier
+            if len(pos_list) < 5:
+                continue
+
+            # add rule here
+            main_direction = extract_main_direction(pos_list,
+                                                    f_direction)  # y x
+            reference_directin = np.array([0, 1]).reshape([-1, 2])  # y x
+            is_h_angle = abs(np.sum(
+                main_direction * reference_directin)) < math.cos(math.pi / 180 *
+                                                                 70)
+
+            point_yxs = np.array(pos_list)
+            max_y, max_x = np.max(point_yxs, axis=0)
+            min_y, min_x = np.min(point_yxs, axis=0)
+            is_h_len = (max_y - min_y) < 1.5 * (max_x - min_x)
+
+            pos_list_final = []
+            if is_h_len:
+                xs = np.unique(xs)
+                for x in xs:
+                    ys = instance_label_map[:, x].copy().reshape((-1, ))
+                    y = int(np.where(ys == instance_id)[0].mean())
+                    pos_list_final.append((y, x))
+            else:
+                ys = np.unique(ys)
+                for y in ys:
+                    xs = instance_label_map[y, :].copy().reshape((-1, ))
+                    x = int(np.where(xs == instance_id)[0].mean())
+                    pos_list_final.append((y, x))
+
+            pos_list_sorted, _ = sort_with_direction(pos_list_final,
+                                                     f_direction)
+            all_pos_yxs.append(pos_list_sorted)
+
+    # use decoder to filter backgroud points.
+    p_char_maps = p_char_maps.transpose([1, 2, 0])
+    decode_res = ctc_decoder_for_image(
+        all_pos_yxs, logits_map=p_char_maps, keep_blank_in_idxs=True)
+    for decoded_str, keep_yxs_list in decode_res:
+        if is_backbone:
+            keep_yxs_list_with_id = add_id(keep_yxs_list, image_id=image_id)
+            instance_center_pos_yxs.append(keep_yxs_list_with_id)
+        else:
+            end_points_yxs.extend((keep_yxs_list[0], keep_yxs_list[-1]))
+            center_pos_yxs.extend(keep_yxs_list)
+
+    if is_backbone:
+        return instance_center_pos_yxs
+    else:
+        return center_pos_yxs, end_points_yxs
+
+
+def generate_pivot_list_slow(p_score,
+                             p_char_maps,
+                             f_direction,
+                             score_thresh=0.5,
+                             is_backbone=False,
+                             is_curved=True,
+                             image_id=0):
+    """
+    Warp all the function together.
+    """
+    if is_curved:
+        return generate_pivot_list_curved(
+            p_score,
+            p_char_maps,
+            f_direction,
+            score_thresh=score_thresh,
+            is_expand=True,
+            is_backbone=is_backbone,
+            image_id=image_id)
+    else:
+        return generate_pivot_list_horizontal(
+            p_score,
+            p_char_maps,
+            f_direction,
+            score_thresh=score_thresh,
+            is_backbone=is_backbone,
+            image_id=image_id)
+
+
+# for refine module
+def extract_main_direction(pos_list, f_direction):
+    """
+    f_direction: h x w x 2
+    pos_list: [[y, x], [y, x], [y, x] ...]
+    """
+    pos_list = np.array(pos_list)
+    point_direction = f_direction[pos_list[:, 0], pos_list[:, 1]]
+    point_direction = point_direction[:, ::-1]  # x, y -> y, x
+    average_direction = np.mean(point_direction, axis=0, keepdims=True)
+    average_direction = average_direction / (
+        np.linalg.norm(average_direction) + 1e-6)
+    return average_direction
+
+
+def sort_by_direction_with_image_id_deprecated(pos_list, f_direction):
+    """
+    f_direction: h x w x 2
+    pos_list: [[id, y, x], [id, y, x], [id, y, x] ...]
+    """
+    pos_list_full = np.array(pos_list).reshape(-1, 3)
+    pos_list = pos_list_full[:, 1:]
+    point_direction = f_direction[pos_list[:, 0], pos_list[:, 1]]  # x, y
+    point_direction = point_direction[:, ::-1]  # x, y -> y, x
+    average_direction = np.mean(point_direction, axis=0, keepdims=True)
+    pos_proj_leng = np.sum(pos_list * average_direction, axis=1)
+    sorted_list = pos_list_full[np.argsort(pos_proj_leng)].tolist()
+    return sorted_list
+
+
+def sort_by_direction_with_image_id(pos_list, f_direction):
+    """
+    f_direction: h x w x 2
+    pos_list: [[y, x], [y, x], [y, x] ...]
+    """
+
+    def sort_part_with_direction(pos_list_full, point_direction):
+        pos_list_full = np.array(pos_list_full).reshape(-1, 3)
+        pos_list = pos_list_full[:, 1:]
+        point_direction = np.array(point_direction).reshape(-1, 2)
+        average_direction = np.mean(point_direction, axis=0, keepdims=True)
+        pos_proj_leng = np.sum(pos_list * average_direction, axis=1)
+        sorted_list = pos_list_full[np.argsort(pos_proj_leng)].tolist()
+        sorted_direction = point_direction[np.argsort(pos_proj_leng)].tolist()
+        return sorted_list, sorted_direction
+
+    pos_list = np.array(pos_list).reshape(-1, 3)
+    point_direction = f_direction[pos_list[:, 1], pos_list[:, 2]]  # x, y
+    point_direction = point_direction[:, ::-1]  # x, y -> y, x
+    sorted_point, sorted_direction = sort_part_with_direction(pos_list,
+                                                              point_direction)
+
+    point_num = len(sorted_point)
+    if point_num >= 16:
+        middle_num = point_num // 2
+        first_part_point = sorted_point[:middle_num]
+        first_point_direction = sorted_direction[:middle_num]
+        sorted_fist_part_point, sorted_fist_part_direction = sort_part_with_direction(
+            first_part_point, first_point_direction)
+
+        last_part_point = sorted_point[middle_num:]
+        last_point_direction = sorted_direction[middle_num:]
+        sorted_last_part_point, sorted_last_part_direction = sort_part_with_direction(
+            last_part_point, last_point_direction)
+        sorted_point = sorted_fist_part_point + sorted_last_part_point
+        sorted_direction = sorted_fist_part_direction + sorted_last_part_direction
+
+    return sorted_point
+
+
+def generate_pivot_list_tt_inference(p_score,
+                                     p_char_maps,
+                                     f_direction,
+                                     score_thresh=0.5,
+                                     is_backbone=False,
+                                     is_curved=True,
+                                     image_id=0):
+    """
+    return center point and end point of TCL instance; filter with the char maps;
+    """
+    p_score = p_score[0]
+    f_direction = f_direction.transpose(1, 2, 0)
+    p_tcl_map = (p_score > score_thresh) * 1.0
+    skeleton_map = thin(p_tcl_map)
+    instance_count, instance_label_map = cv2.connectedComponents(
+        skeleton_map.astype(np.uint8), connectivity=8)
+
+    # get TCL Instance
+    all_pos_yxs = []
+    if instance_count > 0:
+        for instance_id in range(1, instance_count):
+            pos_list = []
+            ys, xs = np.where(instance_label_map == instance_id)
+            pos_list = list(zip(ys, xs))
+            ### FIX-ME, eliminate outlier
+            if len(pos_list) < 3:
+                continue
+            pos_list_sorted = sort_and_expand_with_direction_v2(
+                pos_list, f_direction, p_tcl_map)
+            pos_list_sorted_with_id = add_id(pos_list_sorted, image_id=image_id)
+            all_pos_yxs.append(pos_list_sorted_with_id)
+    return all_pos_yxs

backend/ppocr/utils/e2e_utils/pgnet_pp_utils.py

@@ -0,0 +1,162 @@
+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+import paddle
+import os
+import sys
+
+__dir__ = os.path.dirname(__file__)
+sys.path.append(__dir__)
+sys.path.append(os.path.join(__dir__, '..'))
+from extract_textpoint_slow import *
+from extract_textpoint_fast import generate_pivot_list_fast, restore_poly
+
+
+class PGNet_PostProcess(object):
+    # two different post-process
+    def __init__(self, character_dict_path, valid_set, score_thresh, outs_dict,
+                 shape_list):
+        self.Lexicon_Table = get_dict(character_dict_path)
+        self.valid_set = valid_set
+        self.score_thresh = score_thresh
+        self.outs_dict = outs_dict
+        self.shape_list = shape_list
+
+    def pg_postprocess_fast(self):
+        p_score = self.outs_dict['f_score']
+        p_border = self.outs_dict['f_border']
+        p_char = self.outs_dict['f_char']
+        p_direction = self.outs_dict['f_direction']
+        if isinstance(p_score, paddle.Tensor):
+            p_score = p_score[0].numpy()
+            p_border = p_border[0].numpy()
+            p_direction = p_direction[0].numpy()
+            p_char = p_char[0].numpy()
+        else:
+            p_score = p_score[0]
+            p_border = p_border[0]
+            p_direction = p_direction[0]
+            p_char = p_char[0]
+
+        src_h, src_w, ratio_h, ratio_w = self.shape_list[0]
+        instance_yxs_list, seq_strs = generate_pivot_list_fast(
+            p_score,
+            p_char,
+            p_direction,
+            self.Lexicon_Table,
+            score_thresh=self.score_thresh)
+        poly_list, keep_str_list = restore_poly(instance_yxs_list, seq_strs,
+                                                p_border, ratio_w, ratio_h,
+                                                src_w, src_h, self.valid_set)
+        data = {
+            'points': poly_list,
+            'texts': keep_str_list,
+        }
+        return data
+
+    def pg_postprocess_slow(self):
+        p_score = self.outs_dict['f_score']
+        p_border = self.outs_dict['f_border']
+        p_char = self.outs_dict['f_char']
+        p_direction = self.outs_dict['f_direction']
+        if isinstance(p_score, paddle.Tensor):
+            p_score = p_score[0].numpy()
+            p_border = p_border[0].numpy()
+            p_direction = p_direction[0].numpy()
+            p_char = p_char[0].numpy()
+        else:
+            p_score = p_score[0]
+            p_border = p_border[0]
+            p_direction = p_direction[0]
+            p_char = p_char[0]
+        src_h, src_w, ratio_h, ratio_w = self.shape_list[0]
+        is_curved = self.valid_set == "totaltext"
+        char_seq_idx_set, instance_yxs_list = generate_pivot_list_slow(
+            p_score,
+            p_char,
+            p_direction,
+            score_thresh=self.score_thresh,
+            is_backbone=True,
+            is_curved=is_curved)
+        seq_strs = []
+        for char_idx_set in char_seq_idx_set:
+            pr_str = ''.join([self.Lexicon_Table[pos] for pos in char_idx_set])
+            seq_strs.append(pr_str)
+        poly_list = []
+        keep_str_list = []
+        all_point_list = []
+        all_point_pair_list = []
+        for yx_center_line, keep_str in zip(instance_yxs_list, seq_strs):
+            if len(yx_center_line) == 1:
+                yx_center_line.append(yx_center_line[-1])
+
+            offset_expand = 1.0
+            if self.valid_set == 'totaltext':
+                offset_expand = 1.2
+
+            point_pair_list = []
+            for batch_id, y, x in yx_center_line:
+                offset = p_border[:, y, x].reshape(2, 2)
+                if offset_expand != 1.0:
+                    offset_length = np.linalg.norm(
+                        offset, axis=1, keepdims=True)
+                    expand_length = np.clip(
+                        offset_length * (offset_expand - 1),
+                        a_min=0.5,
+                        a_max=3.0)
+                    offset_detal = offset / offset_length * expand_length
+                    offset = offset + offset_detal
+                ori_yx = np.array([y, x], dtype=np.float32)
+                point_pair = (ori_yx + offset)[:, ::-1] * 4.0 / np.array(
+                    [ratio_w, ratio_h]).reshape(-1, 2)
+                point_pair_list.append(point_pair)
+
+                all_point_list.append([
+                    int(round(x * 4.0 / ratio_w)),
+                    int(round(y * 4.0 / ratio_h))
+                ])
+                all_point_pair_list.append(point_pair.round().astype(np.int32)
+                                           .tolist())
+
+            detected_poly, pair_length_info = point_pair2poly(point_pair_list)
+            detected_poly = expand_poly_along_width(
+                detected_poly, shrink_ratio_of_width=0.2)
+            detected_poly[:, 0] = np.clip(
+                detected_poly[:, 0], a_min=0, a_max=src_w)
+            detected_poly[:, 1] = np.clip(
+                detected_poly[:, 1], a_min=0, a_max=src_h)
+
+            if len(keep_str) < 2:
+                continue
+
+            keep_str_list.append(keep_str)
+            detected_poly = np.round(detected_poly).astype('int32')
+            if self.valid_set == 'partvgg':
+                middle_point = len(detected_poly) // 2
+                detected_poly = detected_poly[
+                    [0, middle_point - 1, middle_point, -1], :]
+                poly_list.append(detected_poly)
+            elif self.valid_set == 'totaltext':
+                poly_list.append(detected_poly)
+            else:
+                print('--> Not supported format.')
+                exit(-1)
+        data = {
+            'points': poly_list,
+            'texts': keep_str_list,
+        }
+        return data

backend/ppocr/utils/e2e_utils/visual.py

@@ -0,0 +1,162 @@
+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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 numpy as np
+import cv2
+import time
+
+
+def resize_image(im, max_side_len=512):
+    """
+    resize image to a size multiple of max_stride which is required by the network
+    :param im: the resized image
+    :param max_side_len: limit of max image size to avoid out of memory in gpu
+    :return: the resized image and the resize ratio
+    """
+    h, w, _ = im.shape
+
+    resize_w = w
+    resize_h = h
+
+    if resize_h > resize_w:
+        ratio = float(max_side_len) / resize_h
+    else:
+        ratio = float(max_side_len) / resize_w
+
+    resize_h = int(resize_h * ratio)
+    resize_w = int(resize_w * ratio)
+
+    max_stride = 128
+    resize_h = (resize_h + max_stride - 1) // max_stride * max_stride
+    resize_w = (resize_w + max_stride - 1) // max_stride * max_stride
+    im = cv2.resize(im, (int(resize_w), int(resize_h)))
+    ratio_h = resize_h / float(h)
+    ratio_w = resize_w / float(w)
+
+    return im, (ratio_h, ratio_w)
+
+
+def resize_image_min(im, max_side_len=512):
+    """
+    """
+    h, w, _ = im.shape
+
+    resize_w = w
+    resize_h = h
+
+    if resize_h < resize_w:
+        ratio = float(max_side_len) / resize_h
+    else:
+        ratio = float(max_side_len) / resize_w
+
+    resize_h = int(resize_h * ratio)
+    resize_w = int(resize_w * ratio)
+
+    max_stride = 128
+    resize_h = (resize_h + max_stride - 1) // max_stride * max_stride
+    resize_w = (resize_w + max_stride - 1) // max_stride * max_stride
+    im = cv2.resize(im, (int(resize_w), int(resize_h)))
+    ratio_h = resize_h / float(h)
+    ratio_w = resize_w / float(w)
+    return im, (ratio_h, ratio_w)
+
+
+def resize_image_for_totaltext(im, max_side_len=512):
+    """
+    """
+    h, w, _ = im.shape
+
+    resize_w = w
+    resize_h = h
+    ratio = 1.25
+    if h * ratio > max_side_len:
+        ratio = float(max_side_len) / resize_h
+
+    resize_h = int(resize_h * ratio)
+    resize_w = int(resize_w * ratio)
+
+    max_stride = 128
+    resize_h = (resize_h + max_stride - 1) // max_stride * max_stride
+    resize_w = (resize_w + max_stride - 1) // max_stride * max_stride
+    im = cv2.resize(im, (int(resize_w), int(resize_h)))
+    ratio_h = resize_h / float(h)
+    ratio_w = resize_w / float(w)
+    return im, (ratio_h, ratio_w)
+
+
+def point_pair2poly(point_pair_list):
+    """
+    Transfer vertical point_pairs into poly point in clockwise.
+    """
+    pair_length_list = []
+    for point_pair in point_pair_list:
+        pair_length = np.linalg.norm(point_pair[0] - point_pair[1])
+        pair_length_list.append(pair_length)
+    pair_length_list = np.array(pair_length_list)
+    pair_info = (pair_length_list.max(), pair_length_list.min(),
+                 pair_length_list.mean())
+
+    point_num = len(point_pair_list) * 2
+    point_list = [0] * point_num
+    for idx, point_pair in enumerate(point_pair_list):
+        point_list[idx] = point_pair[0]
+        point_list[point_num - 1 - idx] = point_pair[1]
+    return np.array(point_list).reshape(-1, 2), pair_info
+
+
+def shrink_quad_along_width(quad, begin_width_ratio=0., end_width_ratio=1.):
+    """
+    Generate shrink_quad_along_width.
+    """
+    ratio_pair = np.array(
+        [[begin_width_ratio], [end_width_ratio]], dtype=np.float32)
+    p0_1 = quad[0] + (quad[1] - quad[0]) * ratio_pair
+    p3_2 = quad[3] + (quad[2] - quad[3]) * ratio_pair
+    return np.array([p0_1[0], p0_1[1], p3_2[1], p3_2[0]])
+
+
+def expand_poly_along_width(poly, shrink_ratio_of_width=0.3):
+    """
+    expand poly along width.
+    """
+    point_num = poly.shape[0]
+    left_quad = np.array(
+        [poly[0], poly[1], poly[-2], poly[-1]], dtype=np.float32)
+    left_ratio = -shrink_ratio_of_width * np.linalg.norm(left_quad[0] - left_quad[3]) / \
+                 (np.linalg.norm(left_quad[0] - left_quad[1]) + 1e-6)
+    left_quad_expand = shrink_quad_along_width(left_quad, left_ratio, 1.0)
+    right_quad = np.array(
+        [
+            poly[point_num // 2 - 2], poly[point_num // 2 - 1],
+            poly[point_num // 2], poly[point_num // 2 + 1]
+        ],
+        dtype=np.float32)
+    right_ratio = 1.0 + \
+                  shrink_ratio_of_width * np.linalg.norm(right_quad[0] - right_quad[3]) / \
+                  (np.linalg.norm(right_quad[0] - right_quad[1]) + 1e-6)
+    right_quad_expand = shrink_quad_along_width(right_quad, 0.0, right_ratio)
+    poly[0] = left_quad_expand[0]
+    poly[-1] = left_quad_expand[-1]
+    poly[point_num // 2 - 1] = right_quad_expand[1]
+    poly[point_num // 2] = right_quad_expand[2]
+    return poly
+
+
+def norm2(x, axis=None):
+    if axis:
+        return np.sqrt(np.sum(x**2, axis=axis))
+    return np.sqrt(np.sum(x**2))
+
+
+def cos(p1, p2):
+    return (p1 * p2).sum() / (norm2(p1) * norm2(p2))

backend/ppocr/utils/iou.py

@@ -0,0 +1,54 @@
+# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+"""
+This code is refer from:
+https://github.com/whai362/PSENet/blob/python3/models/loss/iou.py
+"""
+
+import paddle
+
+EPS = 1e-6
+
+
+def iou_single(a, b, mask, n_class):
+    valid = mask == 1
+    a = a.masked_select(valid)
+    b = b.masked_select(valid)
+    miou = []
+    for i in range(n_class):
+        if a.shape == [0] and a.shape == b.shape:
+            inter = paddle.to_tensor(0.0)
+            union = paddle.to_tensor(0.0)
+        else:
+            inter = ((a == i).logical_and(b == i)).astype('float32')
+            union = ((a == i).logical_or(b == i)).astype('float32')
+        miou.append(paddle.sum(inter) / (paddle.sum(union) + EPS))
+    miou = sum(miou) / len(miou)
+    return miou
+
+
+def iou(a, b, mask, n_class=2, reduce=True):
+    batch_size = a.shape[0]
+
+    a = a.reshape([batch_size, -1])
+    b = b.reshape([batch_size, -1])
+    mask = mask.reshape([batch_size, -1])
+
+    iou = paddle.zeros((batch_size, ), dtype='float32')
+    for i in range(batch_size):
+        iou[i] = iou_single(a[i], b[i], mask[i], n_class)
+
+    if reduce:
+        iou = paddle.mean(iou)
+    return iou

backend/ppocr/utils/loggers/__init__.py

@@ -0,0 +1,3 @@
+from .vdl_logger import VDLLogger
+from .wandb_logger import WandbLogger
+from .loggers import Loggers

backend/ppocr/utils/loggers/base_logger.py

@@ -0,0 +1,15 @@
+import os
+from abc import ABC, abstractmethod
+
+class BaseLogger(ABC):
+    def __init__(self, save_dir):
+        self.save_dir = save_dir
+        os.makedirs(self.save_dir, exist_ok=True)
+
+    @abstractmethod
+    def log_metrics(self, metrics, prefix=None):
+        pass
+
+    @abstractmethod
+    def close(self):
+        pass

backend/ppocr/utils/loggers/loggers.py

@@ -0,0 +1,18 @@
+from .wandb_logger import WandbLogger
+
+class Loggers(object):
+    def __init__(self, loggers):
+        super().__init__()
+        self.loggers = loggers
+
+    def log_metrics(self, metrics, prefix=None, step=None):
+        for logger in self.loggers:
+            logger.log_metrics(metrics, prefix=prefix, step=step)
+    
+    def log_model(self, is_best, prefix, metadata=None):
+        for logger in self.loggers:
+            logger.log_model(is_best=is_best, prefix=prefix, metadata=metadata)
+    
+    def close(self):
+        for logger in self.loggers:
+            logger.close()

backend/ppocr/utils/loggers/vdl_logger.py

@@ -0,0 +1,21 @@
+from .base_logger import BaseLogger
+from visualdl import LogWriter
+
+class VDLLogger(BaseLogger):
+    def __init__(self, save_dir):
+        super().__init__(save_dir)
+        self.vdl_writer = LogWriter(logdir=save_dir)
+
+    def log_metrics(self, metrics, prefix=None, step=None):
+        if not prefix:
+            prefix = ""
+        updated_metrics = {prefix + "/" + k: v for k, v in metrics.items()}
+
+        for k, v in updated_metrics.items():
+            self.vdl_writer.add_scalar(k, v, step)
+    
+    def log_model(self, is_best, prefix, metadata=None):
+        pass
+    
+    def close(self):
+        self.vdl_writer.close() 

backend/ppocr/utils/loggers/wandb_logger.py

@@ -0,0 +1,78 @@
+import os
+from .base_logger import BaseLogger
+
+class WandbLogger(BaseLogger):
+    def __init__(self, 
+        project=None, 
+        name=None, 
+        id=None, 
+        entity=None, 
+        save_dir=None, 
+        config=None,
+        **kwargs):
+        try:
+            import wandb
+            self.wandb = wandb
+        except ModuleNotFoundError:
+            raise ModuleNotFoundError(
+                "Please install wandb using `pip install wandb`"
+                )
+
+        self.project = project
+        self.name = name
+        self.id = id
+        self.save_dir = save_dir
+        self.config = config
+        self.kwargs = kwargs
+        self.entity = entity
+        self._run = None
+        self._wandb_init = dict(
+            project=self.project,
+            name=self.name,
+            id=self.id,
+            entity=self.entity,
+            dir=self.save_dir,
+            resume="allow"
+        )
+        self._wandb_init.update(**kwargs)
+
+        _ = self.run
+
+        if self.config:
+            self.run.settings_config.update(self.config)
+
+    @property
+    def run(self):
+        if self._run is None:
+            if self.wandb.run is not None:
+                logger.info(
+                    "There is a wandb run already in progress "
+                    "and newly created instances of `WandbLogger` will reuse"
+                    " this run. If this is not desired, call `wandb.finish()`"
+                    "before instantiating `WandbLogger`."
+                )
+                self._run = self.wandb.run
+            else:
+                self._run = self.wandb.init(**self._wandb_init)
+        return self._run
+
+    def log_metrics(self, metrics, prefix=None, step=None):
+        if not prefix:
+            prefix = ""
+        updated_metrics = {prefix.lower() + "/" + k: v for k, v in metrics.items()}
+        
+        self.run.log(updated_metrics, step=step)
+
+    def log_model(self, is_best, prefix, metadata=None):
+        model_path = os.path.join(self.save_dir, prefix + '.pdparams')
+        artifact = self.wandb.Artifact('model-{}'.format(self.run.id), type='model', metadata=metadata)
+        artifact.add_file(model_path, name="model_ckpt.pdparams")
+
+        aliases = [prefix]
+        if is_best:
+            aliases.append("best")
+
+        self.run.log_artifact(artifact, aliases=aliases)
+
+    def close(self):
+        self.run.finish()

backend/ppocr/utils/logging.py

@@ -0,0 +1,71 @@
+# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+"""
+This code is refer from:
+https://github.com/WenmuZhou/PytorchOCR/blob/master/torchocr/utils/logging.py
+"""
+
+import os
+import sys
+import logging
+import functools
+import paddle.distributed as dist
+
+logger_initialized = {}
+
+
+@functools.lru_cache()
+def get_logger(name='ppocr', log_file=None, log_level=logging.DEBUG):
+    """Initialize and get a logger by name.
+    If the logger has not been initialized, this method will initialize the
+    logger by adding one or two handlers, otherwise the initialized logger will
+    be directly returned. During initialization, a StreamHandler will always be
+    added. If `log_file` is specified a FileHandler will also be added.
+    Args:
+        name (str): Logger name.
+        log_file (str | None): The log filename. If specified, a FileHandler
+            will be added to the logger.
+        log_level (int): The logger level. Note that only the process of
+            rank 0 is affected, and other processes will set the level to
+            "Error" thus be silent most of the time.
+    Returns:
+        logging.Logger: The expected logger.
+    """
+    logger = logging.getLogger(name)
+    if name in logger_initialized:
+        return logger
+    for logger_name in logger_initialized:
+        if name.startswith(logger_name):
+            return logger
+
+    formatter = logging.Formatter(
+        '[%(asctime)s] %(name)s %(levelname)s: %(message)s',
+        datefmt="%Y/%m/%d %H:%M:%S")
+
+    stream_handler = logging.StreamHandler(stream=sys.stdout)
+    stream_handler.setFormatter(formatter)
+    logger.addHandler(stream_handler)
+    if log_file is not None and dist.get_rank() == 0:
+        log_file_folder = os.path.split(log_file)[0]
+        os.makedirs(log_file_folder, exist_ok=True)
+        file_handler = logging.FileHandler(log_file, 'a')
+        file_handler.setFormatter(formatter)
+        logger.addHandler(file_handler)
+    if dist.get_rank() == 0:
+        logger.setLevel(log_level)
+    else:
+        logger.setLevel(logging.ERROR)
+    logger_initialized[name] = True
+    logger.propagate = False
+    return logger

backend/ppocr/utils/network.py

@@ -0,0 +1,84 @@
+# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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 os
+import sys
+import tarfile
+import requests
+from tqdm import tqdm
+
+from ppocr.utils.logging import get_logger
+
+
+def download_with_progressbar(url, save_path):
+    logger = get_logger()
+    response = requests.get(url, stream=True)
+    if response.status_code == 200:
+        total_size_in_bytes = int(response.headers.get('content-length', 1))
+        block_size = 1024  # 1 Kibibyte
+        progress_bar = tqdm(
+            total=total_size_in_bytes, unit='iB', unit_scale=True)
+        with open(save_path, 'wb') as file:
+            for data in response.iter_content(block_size):
+                progress_bar.update(len(data))
+                file.write(data)
+        progress_bar.close()
+    else:
+        logger.error("Something went wrong while downloading models")
+        sys.exit(0)
+
+
+def maybe_download(model_storage_directory, url):
+    # using custom model
+    tar_file_name_list = [
+        'inference.pdiparams', 'inference.pdiparams.info', 'inference.pdmodel'
+    ]
+    if not os.path.exists(
+            os.path.join(model_storage_directory, 'inference.pdiparams')
+    ) or not os.path.exists(
+            os.path.join(model_storage_directory, 'inference.pdmodel')):
+        assert url.endswith('.tar'), 'Only supports tar compressed package'
+        tmp_path = os.path.join(model_storage_directory, url.split('/')[-1])
+        print('download {} to {}'.format(url, tmp_path))
+        os.makedirs(model_storage_directory, exist_ok=True)
+        download_with_progressbar(url, tmp_path)
+        with tarfile.open(tmp_path, 'r') as tarObj:
+            for member in tarObj.getmembers():
+                filename = None
+                for tar_file_name in tar_file_name_list:
+                    if tar_file_name in member.name:
+                        filename = tar_file_name
+                if filename is None:
+                    continue
+                file = tarObj.extractfile(member)
+                with open(
+                        os.path.join(model_storage_directory, filename),
+                        'wb') as f:
+                    f.write(file.read())
+        os.remove(tmp_path)
+
+
+def is_link(s):
+    return s is not None and s.startswith('http')
+
+
+def confirm_model_dir_url(model_dir, default_model_dir, default_url):
+    url = default_url
+    if model_dir is None or is_link(model_dir):
+        if is_link(model_dir):
+            url = model_dir
+        file_name = url.split('/')[-1][:-4]
+        model_dir = default_model_dir
+        model_dir = os.path.join(model_dir, file_name)
+    return model_dir, url

backend/ppocr/utils/poly_nms.py

@@ -0,0 +1,146 @@
+# copyright (c) 2022 PaddlePaddle Authors. All Rights Reserve.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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 numpy as np
+from shapely.geometry import Polygon
+
+
+def points2polygon(points):
+    """Convert k points to 1 polygon.
+
+    Args:
+        points (ndarray or list): A ndarray or a list of shape (2k)
+            that indicates k points.
+
+    Returns:
+        polygon (Polygon): A polygon object.
+    """
+    if isinstance(points, list):
+        points = np.array(points)
+
+    assert isinstance(points, np.ndarray)
+    assert (points.size % 2 == 0) and (points.size >= 8)
+
+    point_mat = points.reshape([-1, 2])
+    return Polygon(point_mat)
+
+
+def poly_intersection(poly_det, poly_gt, buffer=0.0001):
+    """Calculate the intersection area between two polygon.
+
+    Args:
+        poly_det (Polygon): A polygon predicted by detector.
+        poly_gt (Polygon): A gt polygon.
+
+    Returns:
+        intersection_area (float): The intersection area between two polygons.
+    """
+    assert isinstance(poly_det, Polygon)
+    assert isinstance(poly_gt, Polygon)
+
+    if buffer == 0:
+        poly_inter = poly_det & poly_gt
+    else:
+        poly_inter = poly_det.buffer(buffer) & poly_gt.buffer(buffer)
+    return poly_inter.area, poly_inter
+
+
+def poly_union(poly_det, poly_gt):
+    """Calculate the union area between two polygon.
+
+    Args:
+        poly_det (Polygon): A polygon predicted by detector.
+        poly_gt (Polygon): A gt polygon.
+
+    Returns:
+        union_area (float): The union area between two polygons.
+    """
+    assert isinstance(poly_det, Polygon)
+    assert isinstance(poly_gt, Polygon)
+
+    area_det = poly_det.area
+    area_gt = poly_gt.area
+    area_inters, _ = poly_intersection(poly_det, poly_gt)
+    return area_det + area_gt - area_inters
+
+
+def valid_boundary(x, with_score=True):
+    num = len(x)
+    if num < 8:
+        return False
+    if num % 2 == 0 and (not with_score):
+        return True
+    if num % 2 == 1 and with_score:
+        return True
+
+    return False
+
+
+def boundary_iou(src, target):
+    """Calculate the IOU between two boundaries.
+
+    Args:
+       src (list): Source boundary.
+       target (list): Target boundary.
+
+    Returns:
+       iou (float): The iou between two boundaries.
+    """
+    assert valid_boundary(src, False)
+    assert valid_boundary(target, False)
+    src_poly = points2polygon(src)
+    target_poly = points2polygon(target)
+
+    return poly_iou(src_poly, target_poly)
+
+
+def poly_iou(poly_det, poly_gt):
+    """Calculate the IOU between two polygons.
+
+    Args:
+        poly_det (Polygon): A polygon predicted by detector.
+        poly_gt (Polygon): A gt polygon.
+
+    Returns:
+        iou (float): The IOU between two polygons.
+    """
+    assert isinstance(poly_det, Polygon)
+    assert isinstance(poly_gt, Polygon)
+    area_inters, _ = poly_intersection(poly_det, poly_gt)
+    area_union = poly_union(poly_det, poly_gt)
+    if area_union == 0:
+        return 0.0
+    return area_inters / area_union
+
+
+def poly_nms(polygons, threshold):
+    assert isinstance(polygons, list)
+
+    polygons = np.array(sorted(polygons, key=lambda x: x[-1]))
+
+    keep_poly = []
+    index = [i for i in range(polygons.shape[0])]
+
+    while len(index) > 0:
+        keep_poly.append(polygons[index[-1]].tolist())
+        A = polygons[index[-1]][:-1]
+        index = np.delete(index, -1)
+        iou_list = np.zeros((len(index), ))
+        for i in range(len(index)):
+            B = polygons[index[i]][:-1]
+            iou_list[i] = boundary_iou(A, B)
+        remove_index = np.where(iou_list > threshold)
+        index = np.delete(index, remove_index)
+
+    return keep_poly

backend/ppocr/utils/profiler.py

@@ -0,0 +1,110 @@
+# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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 sys
+import paddle
+
+# A global variable to record the number of calling times for profiler
+# functions. It is used to specify the tracing range of training steps.
+_profiler_step_id = 0
+
+# A global variable to avoid parsing from string every time.
+_profiler_options = None
+
+
+class ProfilerOptions(object):
+    '''
+    Use a string to initialize a ProfilerOptions.
+    The string should be in the format: "key1=value1;key2=value;key3=value3".
+    For example:
+      "profile_path=model.profile"
+      "batch_range=[50, 60]; profile_path=model.profile"
+      "batch_range=[50, 60]; tracer_option=OpDetail; profile_path=model.profile"
+    ProfilerOptions supports following key-value pair:
+      batch_range      - a integer list, e.g. [100, 110].
+      state            - a string, the optional values are 'CPU', 'GPU' or 'All'. 
+      sorted_key       - a string, the optional values are 'calls', 'total',
+                         'max', 'min' or 'ave.
+      tracer_option    - a string, the optional values are 'Default', 'OpDetail',
+                         'AllOpDetail'.
+      profile_path     - a string, the path to save the serialized profile data,
+                         which can be used to generate a timeline.
+      exit_on_finished - a boolean.
+    '''
+
+    def __init__(self, options_str):
+        assert isinstance(options_str, str)
+
+        self._options = {
+            'batch_range': [10, 20],
+            'state': 'All',
+            'sorted_key': 'total',
+            'tracer_option': 'Default',
+            'profile_path': '/tmp/profile',
+            'exit_on_finished': True
+        }
+        self._parse_from_string(options_str)
+
+    def _parse_from_string(self, options_str):
+        for kv in options_str.replace(' ', '').split(';'):
+            key, value = kv.split('=')
+            if key == 'batch_range':
+                value_list = value.replace('[', '').replace(']', '').split(',')
+                value_list = list(map(int, value_list))
+                if len(value_list) >= 2 and value_list[0] >= 0 and value_list[
+                        1] > value_list[0]:
+                    self._options[key] = value_list
+            elif key == 'exit_on_finished':
+                self._options[key] = value.lower() in ("yes", "true", "t", "1")
+            elif key in [
+                    'state', 'sorted_key', 'tracer_option', 'profile_path'
+            ]:
+                self._options[key] = value
+
+    def __getitem__(self, name):
+        if self._options.get(name, None) is None:
+            raise ValueError(
+                "ProfilerOptions does not have an option named %s." % name)
+        return self._options[name]
+
+
+def add_profiler_step(options_str=None):
+    '''
+    Enable the operator-level timing using PaddlePaddle's profiler.
+    The profiler uses a independent variable to count the profiler steps.
+    One call of this function is treated as a profiler step.
+    
+    Args:
+      profiler_options - a string to initialize the ProfilerOptions.
+                         Default is None, and the profiler is disabled.
+    '''
+    if options_str is None:
+        return
+
+    global _profiler_step_id
+    global _profiler_options
+
+    if _profiler_options is None:
+        _profiler_options = ProfilerOptions(options_str)
+
+    if _profiler_step_id == _profiler_options['batch_range'][0]:
+        paddle.utils.profiler.start_profiler(
+            _profiler_options['state'], _profiler_options['tracer_option'])
+    elif _profiler_step_id == _profiler_options['batch_range'][1]:
+        paddle.utils.profiler.stop_profiler(_profiler_options['sorted_key'],
+                                            _profiler_options['profile_path'])
+        if _profiler_options['exit_on_finished']:
+            sys.exit(0)
+
+    _profiler_step_id += 1

backend/ppocr/utils/save_load.py

@@ -0,0 +1,185 @@
+# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import errno
+import os
+import pickle
+import six
+
+import paddle
+
+from ppocr.utils.logging import get_logger
+
+__all__ = ['load_model']
+
+
+def _mkdir_if_not_exist(path, logger):
+    """
+    mkdir if not exists, ignore the exception when multiprocess mkdir together
+    """
+    if not os.path.exists(path):
+        try:
+            os.makedirs(path)
+        except OSError as e:
+            if e.errno == errno.EEXIST and os.path.isdir(path):
+                logger.warning(
+                    'be happy if some process has already created {}'.format(
+                        path))
+            else:
+                raise OSError('Failed to mkdir {}'.format(path))
+
+
+def load_model(config, model, optimizer=None, model_type='det'):
+    """
+    load model from checkpoint or pretrained_model
+    """
+    logger = get_logger()
+    global_config = config['Global']
+    checkpoints = global_config.get('checkpoints')
+    pretrained_model = global_config.get('pretrained_model')
+    best_model_dict = {}
+
+    if model_type == 'vqa':
+        checkpoints = config['Architecture']['Backbone']['checkpoints']
+        # load vqa method metric
+        if checkpoints:
+            if os.path.exists(os.path.join(checkpoints, 'metric.states')):
+                with open(os.path.join(checkpoints, 'metric.states'),
+                          'rb') as f:
+                    states_dict = pickle.load(f) if six.PY2 else pickle.load(
+                        f, encoding='latin1')
+                best_model_dict = states_dict.get('best_model_dict', {})
+                if 'epoch' in states_dict:
+                    best_model_dict['start_epoch'] = states_dict['epoch'] + 1
+            logger.info("resume from {}".format(checkpoints))
+
+            if optimizer is not None:
+                if checkpoints[-1] in ['/', '\\']:
+                    checkpoints = checkpoints[:-1]
+                if os.path.exists(checkpoints + '.pdopt'):
+                    optim_dict = paddle.load(checkpoints + '.pdopt')
+                    optimizer.set_state_dict(optim_dict)
+                else:
+                    logger.warning(
+                        "{}.pdopt is not exists, params of optimizer is not loaded".
+                        format(checkpoints))
+        return best_model_dict
+
+    if checkpoints:
+        if checkpoints.endswith('.pdparams'):
+            checkpoints = checkpoints.replace('.pdparams', '')
+        assert os.path.exists(checkpoints + ".pdparams"), \
+            "The {}.pdparams does not exists!".format(checkpoints)
+
+        # load params from trained model
+        params = paddle.load(checkpoints + '.pdparams')
+        state_dict = model.state_dict()
+        new_state_dict = {}
+        for key, value in state_dict.items():
+            if key not in params:
+                logger.warning("{} not in loaded params {} !".format(
+                    key, params.keys()))
+                continue
+            pre_value = params[key]
+            if list(value.shape) == list(pre_value.shape):
+                new_state_dict[key] = pre_value
+            else:
+                logger.warning(
+                    "The shape of model params {} {} not matched with loaded params shape {} !".
+                    format(key, value.shape, pre_value.shape))
+        model.set_state_dict(new_state_dict)
+
+        if optimizer is not None:
+            if os.path.exists(checkpoints + '.pdopt'):
+                optim_dict = paddle.load(checkpoints + '.pdopt')
+                optimizer.set_state_dict(optim_dict)
+            else:
+                logger.warning(
+                    "{}.pdopt is not exists, params of optimizer is not loaded".
+                    format(checkpoints))
+
+        if os.path.exists(checkpoints + '.states'):
+            with open(checkpoints + '.states', 'rb') as f:
+                states_dict = pickle.load(f) if six.PY2 else pickle.load(
+                    f, encoding='latin1')
+            best_model_dict = states_dict.get('best_model_dict', {})
+            if 'epoch' in states_dict:
+                best_model_dict['start_epoch'] = states_dict['epoch'] + 1
+        logger.info("resume from {}".format(checkpoints))
+    elif pretrained_model:
+        load_pretrained_params(model, pretrained_model)
+    else:
+        logger.info('train from scratch')
+    return best_model_dict
+
+
+def load_pretrained_params(model, path):
+    logger = get_logger()
+    if path.endswith('.pdparams'):
+        path = path.replace('.pdparams', '')
+    assert os.path.exists(path + ".pdparams"), \
+        "The {}.pdparams does not exists!".format(path)
+
+    params = paddle.load(path + '.pdparams')
+    state_dict = model.state_dict()
+    new_state_dict = {}
+    for k1 in params.keys():
+        if k1 not in state_dict.keys():
+            logger.warning("The pretrained params {} not in model".format(k1))
+        else:
+            if list(state_dict[k1].shape) == list(params[k1].shape):
+                new_state_dict[k1] = params[k1]
+            else:
+                logger.warning(
+                    "The shape of model params {} {} not matched with loaded params {} {} !".
+                    format(k1, state_dict[k1].shape, k1, params[k1].shape))
+    model.set_state_dict(new_state_dict)
+    logger.info("load pretrain successful from {}".format(path))
+    return model
+
+
+def save_model(model,
+               optimizer,
+               model_path,
+               logger,
+               config,
+               is_best=False,
+               prefix='ppocr',
+               **kwargs):
+    """
+    save model to the target path
+    """
+    _mkdir_if_not_exist(model_path, logger)
+    model_prefix = os.path.join(model_path, prefix)
+    paddle.save(optimizer.state_dict(), model_prefix + '.pdopt')
+    if config['Architecture']["model_type"] != 'vqa':
+        paddle.save(model.state_dict(), model_prefix + '.pdparams')
+        metric_prefix = model_prefix
+    else:
+        if config['Global']['distributed']:
+            model._layers.backbone.model.save_pretrained(model_prefix)
+        else:
+            model.backbone.model.save_pretrained(model_prefix)
+        metric_prefix = os.path.join(model_prefix, 'metric')
+    # save metric and config
+    if is_best:
+        with open(metric_prefix + '.states', 'wb') as f:
+            pickle.dump(kwargs, f, protocol=2)
+        logger.info('save best model is to {}'.format(model_prefix))
+    else:
+        logger.info("save model in {}".format(model_prefix))

backend/ppocr/utils/stats.py

@@ -0,0 +1,72 @@
+# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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 collections
+import numpy as np
+import datetime
+
+__all__ = ['TrainingStats', 'Time']
+
+
+class SmoothedValue(object):
+    """Track a series of values and provide access to smoothed values over a
+    window or the global series average.
+    """
+
+    def __init__(self, window_size):
+        self.deque = collections.deque(maxlen=window_size)
+
+    def add_value(self, value):
+        self.deque.append(value)
+
+    def get_median_value(self):
+        return np.median(self.deque)
+
+
+def Time():
+    return datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S.%f')
+
+
+class TrainingStats(object):
+    def __init__(self, window_size, stats_keys):
+        self.window_size = window_size
+        self.smoothed_losses_and_metrics = {
+            key: SmoothedValue(window_size)
+            for key in stats_keys
+        }
+
+    def update(self, stats):
+        for k, v in stats.items():
+            if k not in self.smoothed_losses_and_metrics:
+                self.smoothed_losses_and_metrics[k] = SmoothedValue(
+                    self.window_size)
+            self.smoothed_losses_and_metrics[k].add_value(v)
+
+    def get(self, extras=None):
+        stats = collections.OrderedDict()
+        if extras:
+            for k, v in extras.items():
+                stats[k] = v
+        for k, v in self.smoothed_losses_and_metrics.items():
+            stats[k] = round(v.get_median_value(), 6)
+
+        return stats
+
+    def log(self, extras=None):
+        d = self.get(extras)
+        strs = []
+        for k, v in d.items():
+            strs.append('{}: {:x<6f}'.format(k, v))
+        strs = ', '.join(strs)
+        return strs

backend/ppocr/utils/utility.py

@@ -0,0 +1,131 @@
+# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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 logging
+import os
+import imghdr
+import cv2
+import random
+import numpy as np
+import paddle
+
+
+def print_dict(d, logger, delimiter=0):
+    """
+    Recursively visualize a dict and
+    indenting acrrording by the relationship of keys.
+    """
+    for k, v in sorted(d.items()):
+        if isinstance(v, dict):
+            logger.info("{}{} : ".format(delimiter * " ", str(k)))
+            print_dict(v, logger, delimiter + 4)
+        elif isinstance(v, list) and len(v) >= 1 and isinstance(v[0], dict):
+            logger.info("{}{} : ".format(delimiter * " ", str(k)))
+            for value in v:
+                print_dict(value, logger, delimiter + 4)
+        else:
+            logger.info("{}{} : {}".format(delimiter * " ", k, v))
+
+
+def get_check_global_params(mode):
+    check_params = ['use_gpu', 'max_text_length', 'image_shape', \
+                    'image_shape', 'character_type', 'loss_type']
+    if mode == "train_eval":
+        check_params = check_params + [ \
+            'train_batch_size_per_card', 'test_batch_size_per_card']
+    elif mode == "test":
+        check_params = check_params + ['test_batch_size_per_card']
+    return check_params
+
+
+def _check_image_file(path):
+    img_end = {'jpg', 'bmp', 'png', 'jpeg', 'rgb', 'tif', 'tiff', 'gif'}
+    return any([path.lower().endswith(e) for e in img_end])
+
+
+def get_image_file_list(img_file):
+    imgs_lists = []
+    if img_file is None or not os.path.exists(img_file):
+        raise Exception("not found any img file in {}".format(img_file))
+
+    img_end = {'jpg', 'bmp', 'png', 'jpeg', 'rgb', 'tif', 'tiff', 'gif'}
+    if os.path.isfile(img_file) and _check_image_file(img_file):
+        imgs_lists.append(img_file)
+    elif os.path.isdir(img_file):
+        for single_file in os.listdir(img_file):
+            file_path = os.path.join(img_file, single_file)
+            if os.path.isfile(file_path) and _check_image_file(file_path):
+                imgs_lists.append(file_path)
+    if len(imgs_lists) == 0:
+        raise Exception("not found any img file in {}".format(img_file))
+    imgs_lists = sorted(imgs_lists)
+    return imgs_lists
+
+
+def check_and_read_gif(img_path):
+    if os.path.basename(img_path)[-3:] in ['gif', 'GIF']:
+        gif = cv2.VideoCapture(img_path)
+        ret, frame = gif.read()
+        if not ret:
+            logger = logging.getLogger('ppocr')
+            logger.info("Cannot read {}. This gif image maybe corrupted.")
+            return None, False
+        if len(frame.shape) == 2 or frame.shape[-1] == 1:
+            frame = cv2.cvtColor(frame, cv2.COLOR_GRAY2RGB)
+        imgvalue = frame[:, :, ::-1]
+        return imgvalue, True
+    return None, False
+
+
+def load_vqa_bio_label_maps(label_map_path):
+    with open(label_map_path, "r", encoding='utf-8') as fin:
+        lines = fin.readlines()
+    lines = [line.strip() for line in lines]
+    if "O" not in lines:
+        lines.insert(0, "O")
+    labels = []
+    for line in lines:
+        if line == "O":
+            labels.append("O")
+        else:
+            labels.append("B-" + line)
+            labels.append("I-" + line)
+    label2id_map = {label: idx for idx, label in enumerate(labels)}
+    id2label_map = {idx: label for idx, label in enumerate(labels)}
+    return label2id_map, id2label_map
+
+
+def set_seed(seed=1024):
+    random.seed(seed)
+    np.random.seed(seed)
+    paddle.seed(seed)
+
+
+class AverageMeter:
+    def __init__(self):
+        self.reset()
+
+    def reset(self):
+        """reset"""
+        self.val = 0
+        self.avg = 0
+        self.sum = 0
+        self.count = 0
+
+    def update(self, val, n=1):
+        """update"""
+        self.val = val
+        self.sum += val * n
+        self.count += n
+        self.avg = self.sum / self.count

backend/ppocr/utils/visual.py

@@ -0,0 +1,98 @@
+# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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 os
+import numpy as np
+from PIL import Image, ImageDraw, ImageFont
+
+
+def draw_ser_results(image,
+                     ocr_results,
+                     font_path="doc/fonts/simfang.ttf",
+                     font_size=18):
+    np.random.seed(2021)
+    color = (np.random.permutation(range(255)),
+             np.random.permutation(range(255)),
+             np.random.permutation(range(255)))
+    color_map = {
+        idx: (color[0][idx], color[1][idx], color[2][idx])
+        for idx in range(1, 255)
+    }
+    if isinstance(image, np.ndarray):
+        image = Image.fromarray(image)
+    elif isinstance(image, str) and os.path.isfile(image):
+        image = Image.open(image).convert('RGB')
+    img_new = image.copy()
+    draw = ImageDraw.Draw(img_new)
+
+    font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
+    for ocr_info in ocr_results:
+        if ocr_info["pred_id"] not in color_map:
+            continue
+        color = color_map[ocr_info["pred_id"]]
+        text = "{}: {}".format(ocr_info["pred"], ocr_info["text"])
+
+        draw_box_txt(ocr_info["bbox"], text, draw, font, font_size, color)
+
+    img_new = Image.blend(image, img_new, 0.5)
+    return np.array(img_new)
+
+
+def draw_box_txt(bbox, text, draw, font, font_size, color):
+    # draw ocr results outline
+    bbox = ((bbox[0], bbox[1]), (bbox[2], bbox[3]))
+    draw.rectangle(bbox, fill=color)
+
+    # draw ocr results
+    start_y = max(0, bbox[0][1] - font_size)
+    tw = font.getsize(text)[0]
+    draw.rectangle(
+        [(bbox[0][0] + 1, start_y), (bbox[0][0] + tw + 1, start_y + font_size)],
+        fill=(0, 0, 255))
+    draw.text((bbox[0][0] + 1, start_y), text, fill=(255, 255, 255), font=font)
+
+
+def draw_re_results(image,
+                    result,
+                    font_path="doc/fonts/simfang.ttf",
+                    font_size=18):
+    np.random.seed(0)
+    if isinstance(image, np.ndarray):
+        image = Image.fromarray(image)
+    elif isinstance(image, str) and os.path.isfile(image):
+        image = Image.open(image).convert('RGB')
+    img_new = image.copy()
+    draw = ImageDraw.Draw(img_new)
+
+    font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
+    color_head = (0, 0, 255)
+    color_tail = (255, 0, 0)
+    color_line = (0, 255, 0)
+
+    for ocr_info_head, ocr_info_tail in result:
+        draw_box_txt(ocr_info_head["bbox"], ocr_info_head["text"], draw, font,
+                     font_size, color_head)
+        draw_box_txt(ocr_info_tail["bbox"], ocr_info_tail["text"], draw, font,
+                     font_size, color_tail)
+
+        center_head = (
+            (ocr_info_head['bbox'][0] + ocr_info_head['bbox'][2]) // 2,
+            (ocr_info_head['bbox'][1] + ocr_info_head['bbox'][3]) // 2)
+        center_tail = (
+            (ocr_info_tail['bbox'][0] + ocr_info_tail['bbox'][2]) // 2,
+            (ocr_info_tail['bbox'][1] + ocr_info_tail['bbox'][3]) // 2)
+
+        draw.line([center_head, center_tail], fill=color_line, width=5)
+
+    img_new = Image.blend(image, img_new, 0.5)
+    return np.array(img_new)