Add the POPP PyLaia model (#1)

- Add the POPP PyLaia model (5690fb1e3215283bafb8b31b62a6f2f899134e17)


Co-authored-by: Solène Tarride <starride-teklia@users.noreply.huggingface.co>

README.md

@@ -1,3 +1,38 @@
 ---
+library_name: PyLaia
 license: mit
+tags:
+- PyLaia
+- PyTorch
+- Handwritten text recognition
+metrics:
+- CER
+- WER
+language:
+- 'fr'
 ---
+
+# POPP handwritten text recognition
+
+This model performs Handwritten Text Recognition on French census documents.
+
+## Model description
+
+The model was trained using the PyLaia library on the [POPP generic](https://github.com/Shulk97/POPP-datasets/).
+
+For training, text-lines were resized with a fixed height of 128 pixels, keeping the original aspect ratio. 
+
+An external 6-gram character language model can be used to improve recognition. The language model is trained on the text from the POPP training set.
+
+## Evaluation results
+
+The model achieves the following results:
+
+| set   | Language model | CER (%)    | WER (%) | N lines   |
+|:------|:---------------|:----------:|:-------:|----------:|
+| test  | no             | 16.49      |   36.26 |     479   |
+| test  | yes            | 16.09      |   34.52 |     479   |
+
+## How to use
+
+Please refer to the [documentation](https://atr.pages.teklia.com/pylaia/).

language_model.arpa.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7d2dc6ec84ff4302315e61554fe7dfc3a44dc304bb1d83c560d6da3f4582bf6
+size 4391627

lexicon.txt

@@ -0,0 +1,87 @@
+<ctc> <ctc>
+! !
+" "
+& &
+' '
+( (
+) )
+- -
+0 0
+1 1
+2 2
+3 3
+4 4
+5 5
+6 6
+7 7
+8 8
+9 9
+: :
+? ?
+A A
+B B
+C C
+D D
+E E
+F F
+G G
+H H
+I I
+J J
+K K
+L L
+M M
+N N
+O O
+P P
+Q Q
+R R
+S S
+T T
+U U
+V V
+W W
+X X
+Y Y
+Z Z
+a a
+b b
+c c
+d d
+e e
+f f
+g g
+h h
+i i
+j j
+k k
+l l
+m m
+n n
+o o
+p p
+q q
+r r
+s s
+t t
+u u
+v v
+w w
+x x
+y y
+z z
+° °
+à à
+â â
+ç ç
+è è
+é é
+ê ê
+ë ë
+î î
+ï ï
+ô ô
+ù ù
+ü ü
+<unk> <unk>
+<space> <space>

syms.txt

@@ -0,0 +1,87 @@
+<ctc> 0
+! 1
+" 2
+& 3
+' 4
+( 5
+) 6
+- 7
+0 8
+1 9
+2 10
+3 11
+4 12
+5 13
+6 14
+7 15
+8 16
+9 17
+: 18
+? 19
+A 20
+B 21
+C 22
+D 23
+E 24
+F 25
+G 26
+H 27
+I 28
+J 29
+K 30
+L 31
+M 32
+N 33
+O 34
+P 35
+Q 36
+R 37
+S 38
+T 39
+U 40
+V 41
+W 42
+X 43
+Y 44
+Z 45
+a 46
+b 47
+c 48
+d 49
+e 50
+f 51
+g 52
+h 53
+i 54
+j 55
+k 56
+l 57
+m 58
+n 59
+o 60
+p 61
+q 62
+r 63
+s 64
+t 65
+u 66
+v 67
+w 68
+x 69
+y 70
+z 71
+° 72
+à 73
+â 74
+ç 75
+è 76
+é 77
+ê 78
+ë 79
+î 80
+ï 81
+ô 82
+ù 83
+ü 84
+<unk> 85
+<space> 86

tokens.txt

@@ -0,0 +1,87 @@
+<ctc>
+!
+"
+&
+'
+(
+)
+-
+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
+°
+à
+â
+ç
+è
+é
+ê
+ë
+î
+ï
+ô
+ù
+ü
+<unk>
+<space>

weights.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:535712a733c8d5e7406d90e39a8c82c065a127dffa6762690d4960cf9d206d7a
+size 42696796