Find screenshots by what you remember — not by filename.
PixelFind is a local-first, browser-based screenshot search tool. It uses OCR (Tesseract.js) to extract text from your screenshots and lets you search them by content.
Deploy to GitHub Pages — see instructions below.
- 📁 Folder selection — Select any local folder of screenshots
- 🔍 OCR-powered search — Extracts text from every image locally
- 🏷️ Auto-tagging — Detects Python, C++, React, SQL, GitHub, Figma, Networking, etc.
- 📂 Categories — Programming, Education, Project, Design, Website, Error
- 🧠 Smart Search — Natural language queries like "python error last week"
- 💾 Local index — Stores everything in IndexedDB, nothing uploaded
- ⚡ Incremental indexing — Only processes new/changed files on re-scan
- 🔄 Manual refresh — Check for new screenshots with a single click
- 📊 Batch processing — 2 images in parallel, never lags your device
- 🔒 100% private — Nothing leaves your browser
- Fork this repo
- Go to Settings → Pages → Source:
GitHub Actions - Push to
main— GitHub Actions auto-deploys - Open
https://yourusername.github.io/PixelFind/ - Use Chrome or Edge (required for File System Access API)
git clone https://github.com/yourusername/PixelFind.git
cd PixelFind
npm install
npm run devOpen http://localhost:5173/PixelFind/
npm run buildOutput goes to dist/. Deploy the dist/ folder to any static host.
| Feature | Chrome | Edge | Firefox | Safari |
|---|---|---|---|---|
| Folder Selection | ✅ | ✅ | ❌ | ❌ |
| OCR | ✅ | ✅ | ✅ | ✅ |
| IndexedDB | ✅ | ✅ | ✅ | ✅ |
Folder selection requires Chrome or Edge. This is a browser security limitation — only Chrome/Edge support the File System Access API.
- React + Vite + TypeScript
- Tailwind CSS — Styling
- Tesseract.js — Local OCR
- Dexie + IndexedDB — Local database
- Fuse.js — Fuzzy search
- Zustand — State management
- Lucide React — Icons
MIT