TableBank
收藏资源简介:
TableBank是由微软亚洲研究院创建的一个基于图像的表格检测和识别数据集,包含417,234个高质量标注的表格。该数据集通过从互联网上的Word和Latex文档中提取,利用弱监督方法自动生成。TableBank旨在解决现有表格分析模型在实际应用中泛化能力不足的问题,通过大规模数据集支持深度学习模型的发展。数据集涵盖多种文档类型,如商业文档、官方文件、研究论文等,适用于广泛的表格分析任务。
TableBank is an image-based table detection and recognition dataset created by Microsoft Research Asia, containing 417,234 high-quality annotated tables. It is automatically generated via weakly-supervised methods by extracting tables from Word and LaTeX documents sourced from the Internet. TableBank aims to address the insufficient generalization ability of existing table analysis models in real-world applications, and supports the development of deep learning models with its large-scale dataset. The dataset covers diverse document types such as business documents, official files, research papers and more, and is applicable to a broad spectrum of table analysis tasks.

- 1TableBank: A Benchmark Dataset for Table Detection and Recognition微软亚洲研究院 · 2020年



