Form-NLU
收藏资源简介:
Form-NLU是由悉尼大学创建的一个专注于表格文档结构理解和关键信息提取的数据集。该数据集包含857个表格图像,涵盖数字、打印和手写三种类型,旨在解决表格设计者意图与用户填写内容之间的对齐问题。数据集通过精细的标注,包括6000个表格键值对和4000个表格键值对,支持对表格布局和关键信息的深入分析。Form-NLU不仅适用于学术研究,也适用于工业界解决实际的表格数据处理问题。
Form-NLU is a dataset developed by the University of Sydney, focusing on tabular document structure understanding and key information extraction. It comprises 857 table images across three categories: digital, printed, and handwritten, with the goal of addressing the alignment issue between the designer's intended form layout and user-filled content. The dataset provides fine-grained annotations including 6,000 table key-value pairs and 4,000 table key-value pairs, enabling in-depth analysis of table layouts and key information. Form-NLU is suitable for both academic research and industrial applications targeting real-world tabular data processing tasks.

- 1Form-NLU: Dataset for the Form Natural Language Understanding悉尼大学 · 2023年



