ChartRevive: Reconstructing Data Visualizations from Chart Images Using MLLM
3.00T1 sourcearXiv cs.HC
Source record
Published by arXiv cs.HC (T1 source). The original is at https://arxiv.org/abs/2609.27146.
Pipeline notes
The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.
SummaryChartRevive, a mixed-initiative system using multimodal LLMs to extract both underlying data and visual design specifications from static chart images. Authors benchmark five MLLMs across five chart types, finding GPT-5.4 performs best, and present an interactive overlay-based verification interface for refining reconstructions.
Why it mattersPairs a five-model benchmark of data plus design extraction with where numeric and spatial recovery still fails, and grounds the best model in a concrete verification UI.
Cited by
No citations on record.
