hyturing/US_tax_forms_donut

NIST-SD2（美国税务表格）Donut数据集是针对文档理解任务处理的NIST特殊数据库2的版本，适用于Donut架构。该数据集包含5590个注释的文档图像（5031个训练图像，559个测试图像），跨越20个税务表格类别。数据集由美国国家标准与技术研究院（NIST）策划，采用MIT许可。总大小为947.33 MB，包含类别标签（20种税务表格类型）和全文地面真相注释。

The NIST-SD2 (US Tax Forms) Donut Dataset is a processed version of the NIST Special Database 2 for document understanding tasks, formatted for use with the Donut architecture. It contains 5,590 annotated document images (5,031 train, 559 test) across 20 tax form classes. The dataset is curated by the National Institute of Standards and Technology (NIST), licensed under MIT, with a total size of 947.33 MB. Annotations include class labels (20 tax form types) and full text ground truth.