Datasets:
dataset_name string | dataset_version string | dataset_split string | data list |
|---|---|---|---|
SP-DocVQA | 1.0 | train | [{"questionId":337,"question":"what is the date mentioned in this letter?","question_types":["handwr(...TRUNCATED) |
DocVQA (Task 1: Single Page Document VQA) Archive
- Source Portal: Robust Reading Competition (RRC) Challenge 17
- Host: Computer Vision Center (CVC) / IIIT Hyderabad (UCSF Industry Documents Library)
- Terms: Hosted purely for non-commercial research and educational purposes.
Repository Contents
images/:spdocvqa_images.tar.gz(~12,767 document images)annotations/:spdocvqa_qas.zip(~50,000 QA pairs across train, val, and test)ocr/:spdocvqa_ocr.tar.gz(Microsoft OCR token representations)imdb/:spdocvqa_imdb.zip(Preprocessed Image Database files for MP-DocVQA framework)
- Downloads last month
- 336