Metraipolis › archive › dc_35e101a58582

SciFigure2Code: An AI-Reconstructed Benchmark for Scientific Figure-to-Code — arXiv:2609.08155

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.08155
License (as archived)
CC0-1.0
Archived
2026-10-16T15:12:05.346Z
SHA-256 of the archived text
552cbf4cd0643accf6aab1f636efc2c09ae09b26447c9e219a3ae3d56740bfef
Stable URL of this page
https://metraipolis.com/documento/dc_35e101a58582

How to cite

SciFigure2Code: An AI-Reconstructed Benchmark for Scientific Figure-to-Code — arXiv:2609.08155 — arXiv (en), https://arxiv.org/abs/2609.08155 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_35e101a58582 (Metraipolis archive dc_35e101a58582, sha256 552cbf4cd0643accf6aab1f636efc2c09ae09b26447c9e219a3ae3d56740bfef)
BibTeX
@misc{dc35e101a58582,
  title = {SciFigure2Code: An AI-Reconstructed Benchmark for Scientific Figure-to-Code — arXiv:2609.08155},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.08155},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_35e101a58582; sha256 552cbf4cd0643accf6aab1f636efc2c09ae09b26447c9e219a3ae3d56740bfef},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.08155. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Scientific figures are the interface through which research claims are inspected and reused, but final published panels rarely expose the data or plotting code that produced them. Recovering this hidden provenance from pixels is therefore underdetermined. We introduce SciFigure2Code, an AI-reconstructed benchmark that instead evaluates presentation recovery: generating editable Python programs that preserve how a scientific panel is arranged and read. Role-specialized Codex agents generate, execute, visually refine, and audit silver-standard presentation programs that capture geometry, visual hierarchy, encodings, annotations, and typography without claiming to recover original measurements or author source code. This reconstruction-and-audit protocol turns final published panels into auditable reference packages; the resulting resource contains 6,740 reviewed panels and SciFigureBench, a balanced 337-panel test set across 31 chart subtypes, five domains, and three complexity levels. Across 14 zero-shot models in image-only and caption-assisted settings, execution, multi-component layouts, axes, legends, and scientific labels remain weak. Claude Opus 4.7 achieves the highest image-only Overall score, Claude Opus 4.6 leads caption-assisted reconstruction, and two-stage plan-then-code prompting improves Overall for all four tested models. SciFigure2Code provides an auditable testbed for agents that construct editable, visually faithful scientific figure presentations.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-09 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.08155 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T14:49:57.281Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).