janvanlooy commited on
Commit
98296c8
1 Parent(s): 5082185

added how to contribute and small changes

Browse files
Files changed (1) hide show
  1. README.md +27 -6
README.md CHANGED
@@ -20,6 +20,7 @@ size_categories:
20
  - [Dataset Summary](#dataset-summary)
21
  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
22
  - [How to use it](#how-to-use-it)
 
23
  - [Dataset Structure](#dataset-structure)
24
  - [Data Instances](#data-instances)
25
  - [Data Fields](#data-fields)
@@ -51,8 +52,8 @@ size_categories:
51
  |v0.1| Release of the Fondant-cc-25m dataset
52
 
53
  ### Dataset Summary
54
- Fondant-cc-25m contains 25 million image URLs available under different [Creative Commons](https://creativecommons.org/)
55
- licenses collected from the [Common Crawl web corpus](https://commoncrawl.org/).
56
  The dataset was created using [Fondant](https://fondant.ai), an open source framework that aims to simplify and speed up
57
  large-scale data processing by making self-contained pipeline components reusable across pipelines, infrastructures and shareable within the community.
58
 
@@ -62,7 +63,7 @@ This dataset can be used for training or fine-tuning image generation or compute
62
 
63
  ### How to use it
64
 
65
- We have prepared a sample fondant pipeline that limits the dataset to PNG files and downloads the images.
66
 
67
  To execute the pipeline locally, you must have [docker compose](https://docs.docker.com/compose/),
68
  [Python](https://python.org) >=3.8 and [Git](https://git-scm.com/) installed on your system.
@@ -96,6 +97,24 @@ image data and file names, TODO XXXXXXXXXXXXXXXXXXXXXXXXXXXXX
96
 
97
  To download the full dataset, TODO XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
98
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
99
  ## Dataset Structure
100
  ### Data Instances
101
  Each data instance corresponds to one image. The URL of the image is in the `image_url` feature, and other features (`alt_text`, `webpage_url`, etc) provide some
@@ -128,7 +147,9 @@ fondant-cc-25m is built from CommonCrawl dumps. These dumps are constructed from
128
  Permissive licenses have minimal restrictions on how the image can be copied, modified, and redistributed.
129
  The full list of licenses can be found [here](https://creativecommons.org/about/cclicenses/).
130
  We examined HTML tags of the webpages for the presence of Creative Commons license URLs. A webpage was marked permissive only when a license URL was found in
131
- its footer, aside or sidebar. This was the case in around 0.164% of images.
 
 
132
  Subsequently, all the image URLs present on the web page were collected together with the license information.
133
  A manual test of 1032 randomly sampled images showed an accuracy of 96.32% in which case the image was actually released under a Creative Commons license.
134
  False positives could be due to parsing errors but also incorrect attributions: images indicated by the publisher to be CC which are not.
@@ -158,8 +179,8 @@ As toxic or biased data is prevalent on the internet, it is possible that our da
158
 
159
  ### Dataset Curators
160
  1. Sharon Grundmann, ML6, sharon.grundmann@ml6.eu
161
- 2. Robbe Sneyders, ML6, robbe.sneyders@ml6.eu
162
- 3. Matthias Richter, ML6, matthias.richter@ml6.eu
163
 
164
  ### Licensing Information
165
  Fondant-cc-25m is a collection of images with various Creative Commons and other public licenses. Any use of all or part of the images gathered in Fondant-cc-25m
 
20
  - [Dataset Summary](#dataset-summary)
21
  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
22
  - [How to use it](#how-to-use-it)
23
+ - [How to contribute](#how-to-contribute)
24
  - [Dataset Structure](#dataset-structure)
25
  - [Data Instances](#data-instances)
26
  - [Data Fields](#data-fields)
 
52
  |v0.1| Release of the Fondant-cc-25m dataset
53
 
54
  ### Dataset Summary
55
+ Fondant-cc-25m contains 25 million image URLs with their respective [Creative Commons](https://creativecommons.org/)
56
+ license information collected from the [Common Crawl web corpus](https://commoncrawl.org/).
57
  The dataset was created using [Fondant](https://fondant.ai), an open source framework that aims to simplify and speed up
58
  large-scale data processing by making self-contained pipeline components reusable across pipelines, infrastructures and shareable within the community.
59
 
 
63
 
64
  ### How to use it
65
 
66
+ We have prepared a sample Fondant pipeline for downloading the dataset or part of it.
67
 
68
  To execute the pipeline locally, you must have [docker compose](https://docs.docker.com/compose/),
69
  [Python](https://python.org) >=3.8 and [Git](https://git-scm.com/) installed on your system.
 
97
 
98
  To download the full dataset, TODO XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
99
 
100
+ ### How to contribute
101
+ If you want to contribute to the dataset, the best way is to help us develop pipeline components for further processing. Components
102
+ we are currently looking to add are the following ([GitHub issues](https://github.com/ml6team/fondant/issues?q=is%3Aissue+is%3Aopen+label%3A%22Component+Contribution%22)):
103
+ - Image-based deduplication
104
+ - Visual quality / aesthetic quality estimation
105
+ - Automatic captioning
106
+ - Not safe for work (NSFW) content detection
107
+ - Watermark detection
108
+ - Face detection
109
+ - Personal Identifiable Information (PII) detection
110
+ - Text detection
111
+ - AI generated image detection
112
+ - CLIP embedding generation
113
+ - Image-text CLIP similarity
114
+ - ...
115
+ We are also looking for core framework contributors and users who are willing to give feedback on usability and suggest potential improvements
116
+
117
+
118
  ## Dataset Structure
119
  ### Data Instances
120
  Each data instance corresponds to one image. The URL of the image is in the `image_url` feature, and other features (`alt_text`, `webpage_url`, etc) provide some
 
147
  Permissive licenses have minimal restrictions on how the image can be copied, modified, and redistributed.
148
  The full list of licenses can be found [here](https://creativecommons.org/about/cclicenses/).
149
  We examined HTML tags of the webpages for the presence of Creative Commons license URLs. A webpage was marked permissive only when a license URL was found in
150
+ its footer, aside or sidebar. This was the case only in around 0.164% of images which suggests that image generation models trained on a random sample from
151
+ the public internet may be trained on up to 99.836% images that may be copyrighted.
152
+
153
  Subsequently, all the image URLs present on the web page were collected together with the license information.
154
  A manual test of 1032 randomly sampled images showed an accuracy of 96.32% in which case the image was actually released under a Creative Commons license.
155
  False positives could be due to parsing errors but also incorrect attributions: images indicated by the publisher to be CC which are not.
 
179
 
180
  ### Dataset Curators
181
  1. Sharon Grundmann, ML6, sharon.grundmann@ml6.eu
182
+ 2. Matthias Richter, ML6, matthias.richter@ml6.eu
183
+ 3. Robbe Sneyders, ML6, robbe.sneyders@ml6.eu
184
 
185
  ### Licensing Information
186
  Fondant-cc-25m is a collection of images with various Creative Commons and other public licenses. Any use of all or part of the images gathered in Fondant-cc-25m