Datasets:
Tasks:
Visual Question Answering
Formats:
parquet
Languages:
English
Size:
10K - 100K
ArXiv:
Tags:
medical
License:
flaviagiammarino
commited on
Commit
·
034e40e
1
Parent(s):
8f8814f
Update README.md
Browse files
README.md
CHANGED
@@ -11,6 +11,8 @@ size_categories:
|
|
11 |
- 10K<n<100K
|
12 |
---
|
13 |
|
|
|
|
|
14 |
## Dataset Description
|
15 |
PathVQA is a dataset of question-answer pairs on pathology images. The dataset is intended to be used for training and testing
|
16 |
Medical Visual Question Answering (VQA) systems. The questions contained in the dataset are similar to those in the American
|
@@ -37,5 +39,39 @@ and ranks models based on three metrics: "Yes/No Accuracy", "Free-form accuracy"
|
|
37 |
the accuracy of a model's generated answers for the subset of binary "yes/no" questions. "Free-form accuracy" is the accuracy
|
38 |
of a model's generated answers for the subset of open-ended questions. "Overall accuracy" is the accuracy of a model's generated
|
39 |
answers across all questions.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
|
41 |
|
|
|
11 |
- 10K<n<100K
|
12 |
---
|
13 |
|
14 |
+
# PathVQA
|
15 |
+
|
16 |
## Dataset Description
|
17 |
PathVQA is a dataset of question-answer pairs on pathology images. The dataset is intended to be used for training and testing
|
18 |
Medical Visual Question Answering (VQA) systems. The questions contained in the dataset are similar to those in the American
|
|
|
39 |
the accuracy of a model's generated answers for the subset of binary "yes/no" questions. "Free-form accuracy" is the accuracy
|
40 |
of a model's generated answers for the subset of open-ended questions. "Overall accuracy" is the accuracy of a model's generated
|
41 |
answers across all questions.
|
42 |
+
|
43 |
+
#### Languages
|
44 |
+
The question-answer pairs are in English.
|
45 |
+
|
46 |
+
## Dataset Structure
|
47 |
+
|
48 |
+
### Data Instances
|
49 |
+
Each instance consists of an image-question-answer triplet. The 'image' field is the filename of the referenced image,
|
50 |
+
excluding the file extension, which is '.jpg'.
|
51 |
+
|
52 |
+
Example from training dataset:
|
53 |
+
```python
|
54 |
+
{
|
55 |
+
'image': 'train_0001',
|
56 |
+
'question': 'What is the appearance of the chromatin texture, with fine and coarse clumps?',
|
57 |
+
'answer': 'a salt-and-pepper pattern'
|
58 |
+
}
|
59 |
+
```
|
60 |
+
Example from validation dataset:
|
61 |
+
```python
|
62 |
+
{
|
63 |
+
'image': 'val_0000',
|
64 |
+
'question': 'How are the organisms?',
|
65 |
+
'answer': 'somewhat variable in size'
|
66 |
+
}
|
67 |
+
```
|
68 |
+
Example from test dataset:
|
69 |
+
```python
|
70 |
+
{
|
71 |
+
'image': 'test_0001',
|
72 |
+
'question': 'Is squamous cell carcinoma composed of nests of malignant cells that partially recapitulate the stratified organization of squamous epithelium?',
|
73 |
+
'answer': 'yes'
|
74 |
+
}
|
75 |
+
```
|
76 |
|
77 |
|