Extracts precise numerical data from scientific figures and charts using AI vision and OpenCV calibration.
Scientific Data Extractor is a specialized ClawBio skill designed to digitize data from complex scientific imagery for meta-analyses and systematic reviews. It leverages a sophisticated four-phase pipeline—panel detection, pre-analysis, OpenCV-assisted calibration, and heuristic validation—to convert static images into structured CSV or JSON data. Whether handling Kaplan-Meier curves, forest plots, or bioinformatics-specific visualizations like volcano plots, this skill ensures high-fidelity data recovery by combining Claude's visual reasoning with pixel-level computer vision accuracy.
主要功能
01Hybrid extraction pipeline using Claude Vision and OpenCV pixel-level calibration
02753 GitHub stars
03Multi-format output support for CSV, structured JSON, and interactive Web UI
04Heuristic validation for axis ranges, series counts, and error bar polarity
05Supports 26+ specialized plot types including Kaplan-Meier, forest, and box plots
06Automated multi-panel figure detection and sub-plot splitting
使用场景
01Automating the extraction of numerical values from bioinformatics visualizations like Manhattan plots
02Digitizing published charts for meta-analyses and systematic literature reviews
03Converting legacy research figures into machine-readable datasets for reproduction
