Overview

Chart understanding studies how AI can interpret visualizations, from reading axes, legends, and plotted values to answering questions, comparing charts, and summarizing findings. Common approaches combine vision-language models with OCR, structured chart representations, visual grounding, and tools such as code execution; training often uses synthetic charts alongside real examples. Benchmarks increasingly test not only question answering but also data extraction, chart editing, dynamic interactions, and specialized settings such as scientific and financial figures. Models have improved, but reliable numerical reading, evidence-grounded reasoning, and robustness to dense or unfamiliar charts remain open challenges.

Latest papers

CardsList

63 papers2 papers in the last four weeks, down 71% on the four weeks before. 0.0% of all new papers.

  1. From Charts to Code: A Hierarchical Benchmark for Multimodal Models

    Oct 20, 2025Jiahao Tang, Henry Hengyuan Zhao, Lijian Wu +8Data VisualizationChart-to-Code Generation

  2. ChartHal: A Fine-grained Framework Evaluating Hallucination of Large Vision Language Models in Chart Understanding

    Sep 22, 2025Xingqi Wang, Yiming Cui, Xin Yao +3VLM HallucinationVLM Evaluation

  3. Human-centered evaluation of data visualization understanding in vision-language models

    May 22, 2025Arnav Verma, Kushin Mukherjee, Christopher Potts +2VLM EvaluationChart Understanding