HierDoc

该数据集名为HierDoc，包含了650篇科学论文的图像样本及其内容标签，提供了文档级别和实体级别的注释。这些文档涵盖了物理、数学、计算机科学和定量生物学等8个领域。每篇文档均可在遵循知识共享署名4.0许可协议的前提下自由分发。该数据集的规模包括350张训练图像和300张测试图像，其任务是提取目录内容。This dataset is named HierDoc. It contains 650 image samples of scientific papers along with their content tags, and provides document-level and entity-level annotations. These documents cover 8 disciplines including physics, mathematics, computer science and quantitative biology. All documents can be freely distributed under the Creative Commons Attribution 4.0 International License. The dataset consists of 350 training images and 300 test images, and its target task is table of contents extraction.

搜集汇总

背景与挑战

背景概述

HierDoc数据集是一个包含650篇科学论文图像样本的资源，提供文档和实体级别的注释，覆盖物理、数学等8个领域，用于目录内容提取任务，分为350张训练图像和300张测试图像，遵循知识共享署名4.0许可协议。