Datasets:

Languages:
English
ArXiv:
License:
yasumasaonoe commited on
Commit
92fa3d9
·
verified ·
1 Parent(s): 5477a08

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +10 -16
README.md CHANGED
@@ -100,31 +100,26 @@ DOCCI-AAR | 4,932 | 5,000 | -- | --
100
 
101
  ### Curation Rationale
102
 
103
- [More Information Needed]
104
 
105
  ### Source Data
106
 
107
- #### Initial Data Collection and Normalization
108
-
109
- [More Information Needed]
110
 
111
- #### Who are the source language producers?
112
-
113
- [More Information Needed]
114
 
115
  ### Annotations
116
 
117
  #### Annotation process
118
 
119
- [More Information Needed]
120
-
121
- #### Who are the annotators?
122
-
123
- [More Information Needed]
124
 
125
  ### Personal and Sensitive Information
126
 
127
- [More Information Needed]
 
 
128
 
129
  ## Considerations for Using the Data
130
 
@@ -140,12 +135,11 @@ DOCCI-AAR | 4,932 | 5,000 | -- | --
140
 
141
  [More Information Needed]
142
 
143
-
144
-
145
  ### Licensing Information
146
 
147
  CC BY 4.0
148
 
149
  ### Citation Information
150
 
151
- [More Information Needed]
 
 
100
 
101
  ### Curation Rationale
102
 
103
+ DOCCI is meant to be an evaluation dataset for both text-to-image (T2I) and image-to-text (I2T) generation. See our paper for more details.
104
 
105
  ### Source Data
106
 
107
+ #### Initial Data Collection
 
 
108
 
109
+ All images were taken by one of the authors and their family.
 
 
110
 
111
  ### Annotations
112
 
113
  #### Annotation process
114
 
115
+ All text descriptions were written by human annotators.
116
+ We do not rely on any automated process in our data annotation pipeline.
 
 
 
117
 
118
  ### Personal and Sensitive Information
119
 
120
+ We manually reviewed all images for personally identifiable information (PII), removing some images and blurring detected faces, phone numbers, and URLs to protect privacy.
121
+ For text descriptions, we instructed annotators to exclude any PII, such as people's names, phone numbers, and URLs.
122
+ After the annotation phase, we employed automatic tools to scan for PII, ensuring the descriptions remained free of such information.
123
 
124
  ## Considerations for Using the Data
125
 
 
135
 
136
  [More Information Needed]
137
 
 
 
138
  ### Licensing Information
139
 
140
  CC BY 4.0
141
 
142
  ### Citation Information
143
 
144
+ ```
145
+ ```