Datasets:
Tasks:
Visual Question Answering
Formats:
parquet
Languages:
English
Size:
10K - 100K
ArXiv:
Tags:
medical
License:
flaviagiammarino
commited on
Commit
•
fdcdd3e
1
Parent(s):
68167c7
Update README.md
Browse files
README.md
CHANGED
@@ -47,7 +47,7 @@ publishers and authors of these two books, and the owners of the PEIR digital li
|
|
47 |
**Leaderboard:** [Papers with Code Leaderboard](https://paperswithcode.com/sota/medical-visual-question-answering-on-pathvqa)
|
48 |
|
49 |
### Dataset Summary
|
50 |
-
The
|
51 |
see the [commit](https://github.com/UCSD-AI4H/PathVQA/commit/117e7f4ef88a0e65b0e7f37b98a73d6237a3ceab)
|
52 |
in the GitHub repository. This version of the dataset contains a total of 5,004 images and 32,795 question-answer pairs.
|
53 |
Out of the 5,004 images, 4,289 images are referenced by a question-answer pair, while 715 images are not used.
|
@@ -55,7 +55,7 @@ There are a few image-question-answer triplets which occur more than once in the
|
|
55 |
After dropping the duplicate image-question-answer triplets, the dataset contains 32,632 question-answer pairs on 4,289 images.
|
56 |
|
57 |
#### Supported Tasks and Leaderboards
|
58 |
-
|
59 |
where models are ranked based on three metrics: "Yes/No Accuracy", "Free-form accuracy" and "Overall accuracy". "Yes/No Accuracy" is
|
60 |
the accuracy of a model's generated answers for the subset of binary "yes/no" questions. "Free-form accuracy" is the accuracy
|
61 |
of a model's generated answers for the subset of open-ended questions. "Overall accuracy" is the accuracy of a model's generated
|
|
|
47 |
**Leaderboard:** [Papers with Code Leaderboard](https://paperswithcode.com/sota/medical-visual-question-answering-on-pathvqa)
|
48 |
|
49 |
### Dataset Summary
|
50 |
+
The dataset was obtained from the updated Google Drive link shared by the authors on Feb 15, 2023,
|
51 |
see the [commit](https://github.com/UCSD-AI4H/PathVQA/commit/117e7f4ef88a0e65b0e7f37b98a73d6237a3ceab)
|
52 |
in the GitHub repository. This version of the dataset contains a total of 5,004 images and 32,795 question-answer pairs.
|
53 |
Out of the 5,004 images, 4,289 images are referenced by a question-answer pair, while 715 images are not used.
|
|
|
55 |
After dropping the duplicate image-question-answer triplets, the dataset contains 32,632 question-answer pairs on 4,289 images.
|
56 |
|
57 |
#### Supported Tasks and Leaderboards
|
58 |
+
The PathVQA dataset has an active leaderboard on [Papers with Code](https://paperswithcode.com/sota/medical-visual-question-answering-on-pathvqa)
|
59 |
where models are ranked based on three metrics: "Yes/No Accuracy", "Free-form accuracy" and "Overall accuracy". "Yes/No Accuracy" is
|
60 |
the accuracy of a model's generated answers for the subset of binary "yes/no" questions. "Free-form accuracy" is the accuracy
|
61 |
of a model's generated answers for the subset of open-ended questions. "Overall accuracy" is the accuracy of a model's generated
|