Update README.md
Browse files
README.md
CHANGED
@@ -4,6 +4,7 @@ task_categories:
|
|
4 |
- visual-question-answering
|
5 |
- image-to-text
|
6 |
language:
|
|
|
7 |
- ar
|
8 |
- de
|
9 |
- vi
|
@@ -16,6 +17,7 @@ language:
|
|
16 |
tags:
|
17 |
- multilingual
|
18 |
- text-centric
|
|
|
19 |
size_categories:
|
20 |
- 10K<n<100K
|
21 |
---
|
@@ -24,8 +26,10 @@ size_categories:
|
|
24 |
The dataset is oriented toward visual question answering of multilingual text scenes in nine languages, including Korean, Japanese, Italian, Russian, Deutsch, French, Thai, Arabic, and Vietnamese. The question-answer pairs are labeled by native annotators following a series of rules. A comprehensive description of the dataset can be found in the paper [MTVQA](https://arxiv.org/pdf/2405.11985).
|
25 |
|
26 |
## - Image Distribution
|
|
|
27 |
<table style="width:60%;">
|
28 |
<tr>
|
|
|
29 |
<td><b>KO</b></td>
|
30 |
<td><b>JA</b></td>
|
31 |
<td><b>IT</b></td>
|
@@ -35,7 +39,7 @@ The dataset is oriented toward visual question answering of multilingual text sc
|
|
35 |
<td><b>TH</b></td>
|
36 |
<td><b>AR</b></td>
|
37 |
<td><b>VI</b></td>
|
38 |
-
<td><b>Total</b
|
39 |
</tr>
|
40 |
<tr>
|
41 |
<td><b>Train Images</b></td>
|
@@ -48,6 +52,7 @@ The dataset is oriented toward visual question answering of multilingual text sc
|
|
48 |
<td>319</td>
|
49 |
<td>568</td>
|
50 |
<td>1139</td>
|
|
|
51 |
</tr>
|
52 |
<tr>
|
53 |
<td><b>Test Images</b></td>
|
@@ -60,6 +65,7 @@ The dataset is oriented toward visual question answering of multilingual text sc
|
|
60 |
<td>116</td>
|
61 |
<td>250</td>
|
62 |
<td>250</td>
|
|
|
63 |
</tr>
|
64 |
<tr>
|
65 |
<td><b>Train QA</b></td>
|
@@ -72,6 +78,7 @@ The dataset is oriented toward visual question answering of multilingual text sc
|
|
72 |
<td>625</td>
|
73 |
<td>1597</td>
|
74 |
<td>4011</td>
|
|
|
75 |
</tr>
|
76 |
<tr>
|
77 |
<td><b>Test QA</b></td>
|
@@ -84,10 +91,10 @@ The dataset is oriented toward visual question answering of multilingual text sc
|
|
84 |
<td>231</td>
|
85 |
<td>703</td>
|
86 |
<td>884</td>
|
|
|
87 |
</tr>
|
88 |
</table>
|
89 |
|
90 |
-
|
91 |
## - LeaderBoard
|
92 |
<table style="width:75%;">
|
93 |
<tr>
|
@@ -349,8 +356,6 @@ from datasets import load_dataset
|
|
349 |
dataset = load_dataset("ByteDance/MTVQA")
|
350 |
```
|
351 |
|
352 |
-
|
353 |
-
|
354 |
## - Out-of-Scope usage
|
355 |
Academic use only, not supported for commercial usage.
|
356 |
|
@@ -373,4 +378,3 @@ Your access to and use of this dataset are at your own risk. We do not guarantee
|
|
373 |
}
|
374 |
```
|
375 |
|
376 |
-
#WIP Coming Soon
|
|
|
4 |
- visual-question-answering
|
5 |
- image-to-text
|
6 |
language:
|
7 |
+
- multilingual
|
8 |
- ar
|
9 |
- de
|
10 |
- vi
|
|
|
17 |
tags:
|
18 |
- multilingual
|
19 |
- text-centric
|
20 |
+
- vqa
|
21 |
size_categories:
|
22 |
- 10K<n<100K
|
23 |
---
|
|
|
26 |
The dataset is oriented toward visual question answering of multilingual text scenes in nine languages, including Korean, Japanese, Italian, Russian, Deutsch, French, Thai, Arabic, and Vietnamese. The question-answer pairs are labeled by native annotators following a series of rules. A comprehensive description of the dataset can be found in the paper [MTVQA](https://arxiv.org/pdf/2405.11985).
|
27 |
|
28 |
## - Image Distribution
|
29 |
+
|
30 |
<table style="width:60%;">
|
31 |
<tr>
|
32 |
+
<td></td>
|
33 |
<td><b>KO</b></td>
|
34 |
<td><b>JA</b></td>
|
35 |
<td><b>IT</b></td>
|
|
|
39 |
<td><b>TH</b></td>
|
40 |
<td><b>AR</b></td>
|
41 |
<td><b>VI</b></td>
|
42 |
+
<td><b>Total</b> </td>
|
43 |
</tr>
|
44 |
<tr>
|
45 |
<td><b>Train Images</b></td>
|
|
|
52 |
<td>319</td>
|
53 |
<td>568</td>
|
54 |
<td>1139</td>
|
55 |
+
<td>6678 </td>
|
56 |
</tr>
|
57 |
<tr>
|
58 |
<td><b>Test Images</b></td>
|
|
|
65 |
<td>116</td>
|
66 |
<td>250</td>
|
67 |
<td>250</td>
|
68 |
+
<td>2116 </td>
|
69 |
</tr>
|
70 |
<tr>
|
71 |
<td><b>Train QA</b></td>
|
|
|
78 |
<td>625</td>
|
79 |
<td>1597</td>
|
80 |
<td>4011</td>
|
81 |
+
<td>21829 </td>
|
82 |
</tr>
|
83 |
<tr>
|
84 |
<td><b>Test QA</b></td>
|
|
|
91 |
<td>231</td>
|
92 |
<td>703</td>
|
93 |
<td>884</td>
|
94 |
+
<td>6778</td>
|
95 |
</tr>
|
96 |
</table>
|
97 |
|
|
|
98 |
## - LeaderBoard
|
99 |
<table style="width:75%;">
|
100 |
<tr>
|
|
|
356 |
dataset = load_dataset("ByteDance/MTVQA")
|
357 |
```
|
358 |
|
|
|
|
|
359 |
## - Out-of-Scope usage
|
360 |
Academic use only, not supported for commercial usage.
|
361 |
|
|
|
378 |
}
|
379 |
```
|
380 |
|
|