OldVis Image Classification Labeler

A web app for classifying old charts and figures. Assign Vis / Map / Text / Table labels (and confidence), filter the catalog, and move through entries quickly. Work stays in your browser — no server database needed.

Live demo

Features

• Classification pairs: Vis / Not Vis, Map / Not Map, Text / Not Text, Table / Not Table, plus Unsure / Confident
• Keyboard shortcuts: 1–0 for labels, A / D for previous / next
• Selectors: filter Unlabeled / Labeled / Unsure and search metadata
• Saved locally: download or upload annotations.json to keep or restore work
• Progress: labeled / unlabeled / unsure / confident counts, with per-category Details

See It In Action

Selectors, image, metadata, label buttons, and progress sit in one screen.

Start Using It

Try the live demo

Open the hosted app:

https://oldvis.github.io/image-classification-labeler/

Run locally

pnpm install
pnpm dev

How Labeling Works

1. Open the app (sample visualizations load automatically).
2. Optionally set your name in the nav (labels still work without it).
3. Click a category button (or press 1–0) to label the current entry.
4. Press Next (D) or Go to First Unlabeled to continue.
5. Download your annotations as JSON, or upload a JSON file to restore work.

For Developers

Command	Description
pnpm install	Install dependencies
pnpm dev	Start the local app (http://localhost:3333)
pnpm build	Build for production (GitHub Pages base path)
pnpm lint	Run ESLint
pnpm typecheck	Check TypeScript types
pnpm test	Run unit tests
pnpm test:e2e	Run end-to-end tests
pnpm docs:screenshot	Update docs/images/screenshot.png for this README

You will need:

• Node.js
• pnpm
• Playwright browsers for e2e / screenshots (pnpm exec playwright install if needed)

Dataset JSON lives under src/assets/ (visualizations.json, annotations.json). The production deploy copies dist/index.html to dist/404.html for SPA deep links on GitHub Pages.

See CONTRIBUTING.md for contribution notes.

This project started from the vitesse-lite template.