Datasets:
image
imagewidth (px) 1.54k
1.54k
| annotation_id
stringclasses 5
values | image_width
int32 1.54k
1.54k
| image_height
int32 2.05k
2.05k
| question
stringclasses 5
values | answers
listlengths 1
4
| answer_bbox
listlengths 4
4
| document_type
stringclasses 1
value | question_type
stringclasses 1
value | language
stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|
3654e497-1688-45ce-a9c4-caa6af357637
| 1,536
| 2,048
|
登録番号は?(番号のみ)
|
[
"T2011001045931"
] |
[
0.3779911976794092,
0.14914054370505148,
0.7824401297609048,
0.1756825048728996
] |
RECEIPT
|
EXTRACTIVE
|
ja
|
|
00bf696a-5a05-4c1e-9d82-006e6b45c43c
| 1,536
| 2,048
|
レシートの日付は?
|
[
"2026/01/22"
] |
[
0.30384222679780176,
0.255308388376444,
0.7217727899486803,
0.29322547575908425
] |
RECEIPT
|
EXTRACTIVE
|
ja
|
|
c0064639-a0f3-4f7e-bdcb-2f3670715576
| 1,536
| 2,048
|
購入品目 (上から最大10) は?
|
[
"デーリーヨーグルッペ200ml",
"レジ袋S",
"高菜明太おにぎり",
"味付海苔海老マヨネーズ"
] |
[
0.266767741356998,
0.3665318446988553,
0.779069721993559,
0.4701718835447385
] |
RECEIPT
|
EXTRACTIVE
|
ja
|
|
b07c4149-1a75-42bf-a5e9-9f000aca8186
| 1,536
| 2,048
|
合計金額 (10%対象) は?
|
[
"3"
] |
[
0.3510279355406429,
0.5194640971421708,
0.7470508482037738,
0.5472699612227736
] |
RECEIPT
|
EXTRACTIVE
|
ja
|
|
1434b218-2ab2-4fdc-b065-0300851784f0
| 1,536
| 2,048
|
合計金額 (8%対象) は?
|
[
"525"
] |
[
0.3695651782610447,
0.5700202136523578,
0.7537916637384655,
0.6028816893839792
] |
RECEIPT
|
EXTRACTIVE
|
ja
|
Business Document VQA Dataset
Visual Question Answering dataset for business document OCR evaluation.
Version: v1.0.1
- Annotations: 5
- Images: 1
- Languages: ja
- Document Types: RECEIPT
Schema
| Field | Type | Description |
|---|---|---|
| image | Image | Document image |
| annotation_id | string | Annotation ID |
| question | string | Question about the document |
| answers | list[string] | Correct answers |
| answer_bbox | list[float] | Bounding box [x0, y0, x1, y1] (0-1 range) |
| document_type | string | Type of business document |
| question_type | string | Category of question |
| language | string | ISO 639-1 language code |
Usage
from datasets import load_dataset
ds = load_dataset("icoxfog417/biz-doc-vqa-test")
print(ds["train"][0])
# Image will be automatically loaded as PIL.Image
- Downloads last month
- 118