Reproduce “VisAnatomy: An SVG Chart Corpus with Fine-Grained Semantic Labels” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Chen Chen, Hannah K. Bako, Peihong Yu, John Hooker, Jeffrey Joyal, Simon C. Wang, Samuel Kim, Jessica Wu and 5 more

Researchers at the University of Maryland introduce VISANATOMY, an SVG chart corpus featuring 942 real-world charts with over 383,000 fine-grained semantic labels. This resource supports advanced AI applications by providing detailed element roles, hierarchical groupings, and visual encodings, demonstrating improved performance in graph neural networks for chart classification and enabling enhanced accessibility features.

Key insights

  • Fine-grained semantic labels, beyond high-level chart types, are crucial for enabling advanced AI applications such as semantic role inference, chart decomposition, and robust classification.
  • Corpus diversity, both in chart designs and originating authoring tools, is essential for developing AI models that generalize effectively to real-world data visualizations.
  • The SVG format's rich structural information facilitates detailed semantic annotation and supports the creation of tool-agnostic AI models for visualization understanding.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.