VisAnatomy
收藏资源简介:
VisAnatomy是一个包含942个真实世界SVG图表的语义标签数据集,由马里兰大学计算机科学系创建。该数据集涵盖40种图表类型,由超过50种工具生成,具有丰富的结构和视觉风格多样性。数据集的创建过程包括手动收集和专家标注,确保了标签的高质量。VisAnatomy主要应用于图表类型分类、图表分解、动画制作和可访问性导航等领域,旨在解决现有图表数据集语义标签不足的问题。
VisAnatomy is a semantic labeling dataset consisting of 942 real-world SVG diagrams, developed by the Department of Computer Science at the University of Maryland. This dataset covers 40 chart types, generated by over 50 different tools, and features rich structural and visual style diversity. The dataset was constructed through manual collection and expert annotation, ensuring high-quality labels. VisAnatomy is primarily applied in fields including chart type classification, chart decomposition, animation production, and accessibility navigation, aiming to address the shortage of semantic labels in existing chart datasets.

- 1VisAnatomy: An SVG Chart Corpus with Fine-Grained Semantic Labels马里兰大学计算机科学系 · 2024年



