Datasets:

Modalities:
Image
Text
Size:
< 1K
ArXiv:
Libraries:
Datasets
License:
davanstrien HF staff commited on
Commit
90e2875
1 Parent(s): 9d3ff07

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +11 -11
README.md CHANGED
@@ -66,10 +66,10 @@ This dataset contains a subset of data used in the paper [You Actually Look Twic
66
 
67
  ## Dataset Structure
68
 
69
- This dataset has two configurations. These configurations both cover the same data and annotations but provide these annotations in different forms to make it easier to intergrate the data with existing processing pipelines.
70
 
71
- - The first configuration `YOLO` uses the original format of the data.
72
- - The second configuration converts the YOLO format into a format which is closer to the `COCO` annotation format. This is done in particular to make it easier to work with the `feature_extractor`s from the `Transformers` models for object detection which expect data to be in a COCO style format.
73
 
74
  ### Data Instances
75
 
@@ -200,21 +200,21 @@ An example instance from the YOLO config:
200
  The fields for the YOLO config:
201
 
202
  - `image`: the image
203
- - `objects`: the annotations which consits of:
204
  - `bbox`: a list of bounding boxes for the image
205
  - `label`: a list of labels for this image
206
 
207
  The fields for the COCO config:
208
 
209
- - `heigh`: height of the image
210
  - `width`: width of the image
211
  - `image`: image
212
  - `image_id`: id for the image
213
  - `objects`: annotations in COCO format, consisting of a list containing dictionaries with the following keys:
214
  - `bbox`: bounding boxes for the images
215
- - `category_id`: label for the image
216
  - `image_id`: id for the image
217
- - `iscrowd`: COCO is crowd flag
218
  - `segmentation`: COCO segmentation annotations (empty in this case but kept for compatibility with other processing scripts)
219
 
220
 
@@ -231,11 +231,11 @@ The dataset contains a train, validation and test split with the following numbe
231
 
232
  ## Dataset Creation
233
 
234
- > [this] dataset was produced using a single source, the Lectaurep Repertoires dataset [Rostaing et al., 2021], which served as a basis for only the training and development split. The testset is composed of original data, from various documents, from the 17th century up to the early 20th with a single soldier war report. The test set is voluntarily very different and out of domainwith column borders that are not drawn nor printed in certain cases, layout in some kind of masonry layout. p.8
235
  .
236
  ### Curation Rationale
237
 
238
- This dataset was created to produce a simplified version of the [Lectaurep Repertoires dataset](https://github.com/HTR-United/lectaurep-repertoires) which was found to contain:
239
 
240
  > around 16 different ways to describe columns, from Col1 to Col7, the case-different col1-col7 and finally ColPair and ColOdd, which we all reduced to Col p.8
241
 
@@ -281,11 +281,11 @@ This data does not contain information relating to living individuals.
281
 
282
  ### Social Impact of Dataset
283
 
284
- There are a growing number of datasets related to page layout for historical documents. This dataset offers a different approach to annotating these datasets (focusing on object detection rather than pixel level annotations). Improving document layout recognition can have a positive impact on downstream tasks, in particular Optical Character Recognition.
285
 
286
  ### Discussion of Biases
287
 
288
- Historical documents contain a wide variety of page layouts. This means that the ability of models trained on this dataset to transfer to documents which may have very different layouts is not guaranteed.
289
 
290
  ### Other Known Limitations
291
 
 
66
 
67
  ## Dataset Structure
68
 
69
+ This dataset has two configurations. These configurations both cover the same data and annotations but provide these annotations in different forms to make it easier to integrate the data with existing processing pipelines.
70
 
71
+ - The first configuration, `YOLO`, uses the data's original format.
72
+ - The second configuration converts the YOLO format into a format which is closer to the `COCO` annotation format. This is done to make it easier to work with the `feature_extractor`s from the `Transformers` models for object detection, which expect data to be in a COCO style format.
73
 
74
  ### Data Instances
75
 
 
200
  The fields for the YOLO config:
201
 
202
  - `image`: the image
203
+ - `objects`: the annotations which consist of:
204
  - `bbox`: a list of bounding boxes for the image
205
  - `label`: a list of labels for this image
206
 
207
  The fields for the COCO config:
208
 
209
+ - `height`: height of the image
210
  - `width`: width of the image
211
  - `image`: image
212
  - `image_id`: id for the image
213
  - `objects`: annotations in COCO format, consisting of a list containing dictionaries with the following keys:
214
  - `bbox`: bounding boxes for the images
215
+ - `category_id`: a label for the image
216
  - `image_id`: id for the image
217
+ - `iscrowd`: COCO `iscrowd` flag
218
  - `segmentation`: COCO segmentation annotations (empty in this case but kept for compatibility with other processing scripts)
219
 
220
 
 
231
 
232
  ## Dataset Creation
233
 
234
+ > [this] dataset was produced using a single source, the Lectaurep Repertoires dataset [Rostaing et al., 2021], which served as a basis for only the training and development split. The testset is composed of original data, from various documents, from the 17th century up to the early 20th with a single soldier war report. The test set is voluntarily very different and out of domain with column borders that are not drawn nor printed in certain cases, layout in some kind of masonry layout. p.8
235
  .
236
  ### Curation Rationale
237
 
238
+ This dataset was created to produce a simplified version of the [Lectaurep Repertoires dataset](https://github.com/HTR-United/lectaurep-repertoires), which was found to contain:
239
 
240
  > around 16 different ways to describe columns, from Col1 to Col7, the case-different col1-col7 and finally ColPair and ColOdd, which we all reduced to Col p.8
241
 
 
281
 
282
  ### Social Impact of Dataset
283
 
284
+ A growing number of datasets are related to page layout for historical documents. This dataset offers a different approach to annotating these datasets (focusing on object detection rather than pixel-level annotations). Improving document layout recognition can have a positive impact on downstream tasks, in particular Optical Character Recognition.
285
 
286
  ### Discussion of Biases
287
 
288
+ Historical documents contain a wide variety of page layouts. This means that the ability of models trained on this dataset to transfer to documents with very different layouts is not guaranteed.
289
 
290
  ### Other Known Limitations
291