babadie yschneider commited on
Commit
1cb898e
1 Parent(s): d13bba3

add-model (#1)

Browse files

- Add files for PyLaia IAM model (9343926ca2eeb8dfcf5f8b47432cc4cfe428583f)
- Link to paper (5254ab2b3f4d56c8d71dbca55282ed3c4d497b22)


Co-authored-by: Schneider <[email protected]>

Files changed (4) hide show
  1. README.md +39 -0
  2. model +0 -0
  3. syms.txt +81 -0
  4. weights.ckpt +3 -0
README.md CHANGED
@@ -1,3 +1,42 @@
1
  ---
 
2
  license: mit
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ library_name: PyLaia
3
  license: mit
4
+ tags:
5
+ - PyLaia
6
+ - PyTorch
7
+ - Handwritten text recognition
8
+ metrics:
9
+ - CER
10
+ - WER
11
+ language:
12
+ - en
13
  ---
14
+
15
+ # English printed text recognition
16
+
17
+ This model performs Handwritten Text Recognition in English.
18
+
19
+ ## Model description
20
+
21
+ The model has been trained using the PyLaia library on the [IAM](https://fki.tic.heia-fr.ch/databases/iam-handwriting-database) dataset.
22
+ Training images were resized with a fixed height of 128 pixels, keeping the original aspect ratio.
23
+
24
+
25
+ ## Evaluation results
26
+
27
+ The model achieves the following results:
28
+
29
+ | Split | CER (%) | WER (%) | Support |
30
+ | ----- | ------- | ------- | ------- |
31
+ | train | 0.32 | 1.26 | 6482 |
32
+ | val | 6.50 | 19.12 | 1926 |
33
+ | test | 7.68 | 19.82 | 1965 |
34
+
35
+ These results were published [Key-value information extraction from full
36
+ handwritten pages](https://arxiv.org/pdf/2304.13530.pdf).
37
+
38
+ Results can be improved by combining PyLaia with a n-gram language model.
39
+
40
+ ## How to use
41
+
42
+ Please refer to the [documentation](https://atr.pages.teklia.com/pylaia/).
model ADDED
Binary file (1.52 kB). View file
 
syms.txt ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <ctc> 0
2
+ ! 1
3
+ " 2
4
+ # 3
5
+ & 4
6
+ ' 5
7
+ ( 6
8
+ ) 7
9
+ * 8
10
+ + 9
11
+ , 10
12
+ - 11
13
+ . 12
14
+ / 13
15
+ 0 14
16
+ 1 15
17
+ 2 16
18
+ 3 17
19
+ 4 18
20
+ 5 19
21
+ 6 20
22
+ 7 21
23
+ 8 22
24
+ 9 23
25
+ : 24
26
+ ; 25
27
+ ? 26
28
+ A 27
29
+ B 28
30
+ C 29
31
+ D 30
32
+ E 31
33
+ F 32
34
+ G 33
35
+ H 34
36
+ I 35
37
+ J 36
38
+ K 37
39
+ L 38
40
+ M 39
41
+ N 40
42
+ O 41
43
+ P 42
44
+ Q 43
45
+ R 44
46
+ S 45
47
+ T 46
48
+ U 47
49
+ V 48
50
+ W 49
51
+ X 50
52
+ Y 51
53
+ Z 52
54
+ a 53
55
+ b 54
56
+ c 55
57
+ d 56
58
+ e 57
59
+ f 58
60
+ g 59
61
+ h 60
62
+ i 61
63
+ j 62
64
+ k 63
65
+ l 64
66
+ m 65
67
+ n 66
68
+ o 67
69
+ p 68
70
+ q 69
71
+ r 70
72
+ s 71
73
+ t 72
74
+ u 73
75
+ v 74
76
+ w 75
77
+ x 76
78
+ y 77
79
+ z 78
80
+ <unk> 79
81
+ <space> 80
weights.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6c5cd4f1157c2b7768fdef5eb0f3264270b477d111f964c7e78a1b18783d09ed
3
+ size 42673218