What it does WebPlotDigitizer lets you upload an image of a chart or graph, calibrate the axes, and extract the underlying numerical data. It runs in a browser and has been kicking around since 2010, accumulating enough academic citations to make its author modestly proud.

The interesting bit The frontend is open source (AGPL v3) and self-hostable, but the “AI Assist” smarts live in closed-source cloud services owned by Automeris LLC. It’s a pragmatic split: the basic digitization works locally; the fancy automation stays behind a sign-up wall.

Key highlights

• Handles “a variety of data visualizations” — exact types are unspecified in the README
• Docker and vanilla npm builds available for local development
• Test suite accessible at localhost:8080/tests
• No official contribution roadmap; consult the maintainer first
• Actively maintained by original author Ankit Rohatgi

Caveats

• The README is thin on technical specifics — no mention of supported chart types, accuracy metrics, or algorithm details
• “AI Assist” and cloud features are closed source; the free tier’s limits are unclear

Verdict Worth bookmarking if you regularly need to scrape data from papers or reports with uncooperative figures. Skip it if you need a fully offline, AI-powered pipeline — the clever bits aren’t in the repo.