Commit
·
e59a709
1
Parent(s):
a34995c
Fixed folder name
Browse files- README.md +2 -2
- images_readme/stats_table.png +3 -0
- images_readme/tasks.png +3 -0
README.md
CHANGED
|
@@ -60,7 +60,7 @@ configs:
|
|
| 60 |
|
| 61 |
# WorldCuisines: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines
|
| 62 |
|
| 63 |
-
 benchmark for multilingual and multicultural understanding through global cuisines. The dataset contains text-image pairs across 30 languages and dialects, spanning 9 language families and featuring over 1 million data points, making it the largest multicultural VQA benchmark as of 17 October 2024.
|
| 66 |
|
|
@@ -81,7 +81,7 @@ From **WC-KB**, we construct **WC-VQA**, a multilingual parallel VQA dataset wit
|
|
| 81 |
We provide **WC-VQA** evaluation datasets in two sizes (12,000 and 60,000 instances) alongside a training dataset (1,080,000 instances).
|
| 82 |
The table below provides more detailed statistics regarding the number of VQA instances and images for each data split.
|
| 83 |
|
| 84 |
-
|
| 64 |
|
| 65 |
WorldCuisines is a massive-scale visual question answering (VQA) benchmark for multilingual and multicultural understanding through global cuisines. The dataset contains text-image pairs across 30 languages and dialects, spanning 9 language families and featuring over 1 million data points, making it the largest multicultural VQA benchmark as of 17 October 2024.
|
| 66 |
|
|
|
|
| 81 |
We provide **WC-VQA** evaluation datasets in two sizes (12,000 and 60,000 instances) alongside a training dataset (1,080,000 instances).
|
| 82 |
The table below provides more detailed statistics regarding the number of VQA instances and images for each data split.
|
| 83 |
|
| 84 |
+
|
| 85 |
|
| 86 |
## Dataset Construction
|
| 87 |
|
images_readme/stats_table.png
ADDED
|
Git LFS Details
|
images_readme/tasks.png
ADDED
|
Git LFS Details
|