Project
figrecover
Public-alpha Python software for recovering approximate source tables from published figures when original data were not released.
figrecover is public-alpha Python software for recovering approximate tabular data from scientific and professional figures when the original source tables were not released.
The workflow combines deterministic PDF and manifest preparation, optional local vision-language-model metadata proposals, calibrated pixel-to-data extraction, auditable CSV/JSON outputs, and review artifacts such as overlays, metrics, and accepted-table manifests.
FRESH uses figrecover for technical-document workflows where figure-derived data need to be traceable before they become modelling inputs, including FEMIC, FHOPS, and other forest-sector modelling pipelines.
Status
- Open software project
People And Roles
- Gregory Paradis: maintainer or lead developer
Outputs
- GitHub repository
- Public documentation
- PyPI package
- Synthetic examples and QA/review workflows
- Integration adapters for FEMIC, FHOPS, and generic modelling workflows