jdsannchao commited on
Commit
7173aa8
1 Parent(s): edf598a

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +28 -0
README.md CHANGED
@@ -1,3 +1,31 @@
1
  ---
2
  license: apache-2.0
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: apache-2.0
3
  ---
4
+
5
+ Here we create two datasets (from existing datasets: CLEVRER, VisualGenome) for the Object Counting instruction tuning task.
6
+
7
+ ### CLEVRER
8
+
9
+ CLEVRER has QA pairs for each 5000 training video
10
+ ```json
11
+ {'video_filename': int, 'scene_index': str (same as filename), 'questions': list [{'question_type': , 'question_subtype': , 'question_text': , 'answer_text': , 'program'(question attributes): }]}
12
+ ```
13
+ We select 'descriptive' type, 'count' subtype questions, they are object counting.
14
+ CLEVRER contains both positive questions and negative (non-exist) questions, so no need to construct negative samples.
15
+ some questions are 'event' specific, counting moving/stationary objects when a certain event happens. i.e., 'How many objects are stationary when the yellow object enters the scene?'
16
+ downloading videos: http://clevrer.csail.mit.edu/
17
+
18
+
19
+ ### VisualGenome
20
+
21
+ We generate some negative questions for non-exist objects in the image. We use the version 1 image sets. Download from: https://homes.cs.washington.edu/~ranjay/visualgenome/api.html
22
+ VisualGenome has 100K+ images. And for the objects in the image, there are attributes, We only focus on the color attributes.
23
+ There are in total 11K+ possible objects. for each image, I add 3 non-exist objects and 1 non-exist attribute for existing objects as negative samples.
24
+ In the original qa dataset, VG has Object Counting questions, we also include them here, with the 'orig_qa'=='Yes'. For those negative questions we generated, 'orig_qa' =='No'.
25
+ ```json
26
+ {'img_id': str, 'orig_qa': Yes/No, 'question_text': 'How many <attribute> <object in plural form> are there? ', 'answer_text': Numbers. or None. }
27
+ ```
28
+ For more details, plz refer to the dataset.
29
+
30
+
31
+