README.md

metadata

language:
  - en

Document VQA Dataset

This dataset contains question-answer pairs for document visual question answering tasks.

Dataset Structure

• image: URL to the document image
• question: Question in English
• answers: Answers in Bengali
• questionId: Unique identifier for each question
• question_types: Type of question (extraction)
• docId: Document identifier
• fonts: Font information

Usage

from datasets import load_dataset

dataset = load_dataset("nihalbaig/docvqa_syn")

Example

You can view the images directly in the dataset viewer on HuggingFace.