mboillet commited on
Commit
d2ec61f
·
verified ·
1 Parent(s): a3d4ac2

Upload dataset

Browse files
Files changed (2) hide show
  1. README.md +4 -75
  2. data/test-00000-of-00001.parquet +2 -2
README.md CHANGED
@@ -1,21 +1,4 @@
1
  ---
2
- task_categories:
3
- - image-to-text
4
- pretty_name: ATR benchmark
5
- size_categories:
6
- - n<1K
7
- language:
8
- - fr
9
- - la
10
- - en
11
- - no
12
- - ar
13
- - zh
14
- - de
15
- tags:
16
- - atr
17
- - htr
18
- - ocr
19
  dataset_info:
20
  features:
21
  - name: dataset
@@ -32,67 +15,13 @@ dataset_info:
32
  dtype: string
33
  splits:
34
  - name: test
35
- num_bytes: 145153440.0
36
- num_examples: 83
37
- download_size: 144199823
38
- dataset_size: 145153440.0
39
  configs:
40
  - config_name: default
41
  data_files:
42
  - split: test
43
  path: data/test-*
44
  ---
45
-
46
- # ATR benchmark - Page/paragraph level
47
-
48
- ## Dataset Description
49
-
50
- - **Homepage:** [ATR benchmark](https://huggingface.co/datasets/Teklia/ATR-benchmark)
51
- - **Point of Contact:** [TEKLIA](https://teklia.com)
52
-
53
- ## Dataset Summary
54
-
55
- The ATR benchmark dataset is a multilingual dataset that includes 83 document images, at page or paragraph level. This dataset has been designed to test ATR models and combines data from several public datasets:
56
- - [BnL Historical Newspapers](https://data.bnl.lu/data/historical-newspapers/)
57
- - [CASIA-HWDB2](https://nlpr.ia.ac.cn/databases/handwriting/Offline_database.html)
58
- - [DIY History - Social Justice](http://diyhistory.lib.uiowa.edu/)
59
- - FINLAM - Historical Newspapers
60
- - [Horae - Books of hours](https://github.com/oriflamms/HORAE)
61
- - [NorHand v3](https://zenodo.org/records/10255840)
62
- - [Marius PELLET](https://europeana.transcribathon.eu/documents/story/?story=121795)
63
- - [RASM](http://www.primaresearch.org/RASM2019/)
64
-
65
- Images are in their original size.
66
-
67
- ### Split
68
-
69
- | dataset | images | language |
70
- | ------------------------------ | ------:| ---------------- |
71
- | BnL Historical Newspapers | 3 | German |
72
- | CASIA-HWDB2 | 10 | Chinese |
73
- | DIY History - Social Justice | 20 | English |
74
- | FINLAM - Historical Newspapers | 10 | English / French |
75
- | Horae - Books of hours | 10 | Latin |
76
- | NorHand v3 | 10 | Norwegian |
77
- | Marius PELLET | 10 | French |
78
- | RASM | 10 | Arabic |
79
-
80
- ## Dataset Structure
81
-
82
- ### Data Instances
83
-
84
- ```
85
- {
86
- 'image': <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size={img size} at 0x1A800E8E190,
87
- 'text': '{transcription}'
88
- }
89
- ```
90
-
91
- ### Data Fields
92
-
93
- - `image`: a PIL.Image.Image object containing the image. Note that when accessing the image column (using dataset[0]["image"]), the image file is automatically decoded. Decoding of a large number of image files might take a significant amount of time. Thus it is important to first query the sample index before the "image" column, i.e. dataset[0]["image"] should always be preferred over dataset["image"][0].
94
- - `dataset`: the name of the original dataset.
95
- - `arkindex_id`: Arkindex element id corresponding to the current page or paragraph.
96
- - `text`: the label transcription of the image.
97
- - `language`: language of text on the image.
98
- - `level`: full document page or a single paragraph.
 
1
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2
  dataset_info:
3
  features:
4
  - name: dataset
 
15
  dtype: string
16
  splits:
17
  - name: test
18
+ num_bytes: 178249701.0
19
+ num_examples: 123
20
+ download_size: 176996271
21
+ dataset_size: 178249701.0
22
  configs:
23
  - config_name: default
24
  data_files:
25
  - split: test
26
  path: data/test-*
27
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
data/test-00000-of-00001.parquet CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3c277e9ceda357ebcaa8ac84c3755662757571cfdd6e96560989613326d8fa6d
3
- size 144199823
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4b98afb72474a82c7e398c735a05b4f61b8a450645340fcc008834967855d15d
3
+ size 176996271