Datasets:

Modalities:
Image
Size:
< 1K
Libraries:
Datasets
License:
Weiyun1025 commited on
Commit
d4cdadb
1 Parent(s): 8320a57

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +1 -0
README.md CHANGED
@@ -7,6 +7,7 @@ The evaluation is formulated as single-choice questions.
7
 
8
  The benchmark consists of four splits:
9
  **Existence**, **Subject**, **Predicate**, and **Object**.
 
10
  The **Existence** split evaluates the object recognition ability while the remaining splits are designed to evaluate the capability of relation comprehension, focusing on probing each of the elements in the subject-predicate-object triplets of the scene graph separately.
11
  Some data examples are shown below.
12
 
 
7
 
8
  The benchmark consists of four splits:
9
  **Existence**, **Subject**, **Predicate**, and **Object**.
10
+
11
  The **Existence** split evaluates the object recognition ability while the remaining splits are designed to evaluate the capability of relation comprehension, focusing on probing each of the elements in the subject-predicate-object triplets of the scene graph separately.
12
  Some data examples are shown below.
13