Dataset Card for EpiCurveBench

EpiCurveBench is a benchmark for chart data extraction, and is made up of 1000 manually curated epidemic curve images collected from diverse sources.

Dataset Details

Dataset Description

Accurate data on disease case counts over time is essential for training reliable disease forecasting models. However, such data is often locked in non-machine-readable formats, most commonly as epidemic curve (epicurve) images—charts that depict case counts over time for a given location. Digitizing these charts would greatly expand the data available for forecasting models, improving their accuracy. Manual digitization, though, is very time-consuming, and existing automated methods struggle with real-world epicurves due to dense data points, overlapping series, and diverse visual styles. To address this, we present EpiCurveBench, a benchmark of 1000 manually curated epicurve images collected from diverse sources. The dataset spans a wide range of chart styles, from simple to highly complex.

Dataset Sources

• Repository: https://github.com/tberkane/EpiCurveBench
• Paper: https://arxiv.org/pdf/2605.27195

Dataset Structure

• images: The 1000 epicurves images.
• tables: TSV files containing ground truth annotations. The second-to-last line in each file contains the minimum axis values for each data row, and the last line contains the corresponding maximum values. These can be used to normalize the extraction performance scores.
• metadata.csv: Metadata for each epicurve image, contains the following columns:
  • id: ID of the corresponding epicurve image.
  • source: type of source the image was collected from.
  • country: country the image reports disease case counts for.
  • years: year/year range the image reports disease case counts for.
  • link: source link.
  • num_series: number of series present in the image.
  • series_len: length of each series on the image.
  • annotation_time: time taken to manually annotate this image
  • chart_type: type of chart.
  • resolution: resolution of the image.

Citation

BibTeX:

@misc{berkane2026epicurvebench,
  title         = {EpiCurveBench: Evaluating VLMs on Epidemic Curve Digitization},
  author        = {Berkane, Thomas and Majumder, Maimuna S.},
  year          = {2026},
  eprint        = {2605.27195},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL}
}