ScanExplain

**Scanned PDFs? We read them for you.**

Every major translation tool — Google Translate, DeepL, Immersive Translate — rejects scanned and image-based PDFs. They need selectable text to work. If your PDF is a photo of a page, they return nothing.

ScanExplain fills this gap.

━━━━━━━━━━━━━━━━━━━━

🔍 WHAT IT DOES

1. In-Browser OCR
Drop any scanned PDF or image. ScanExplain recognizes the text directly in your browser — no file upload, no server processing.

2. AI Translation
Recognized text is translated to Chinese using AI. More target languages coming soon.

3. AI Explanation
Not just translation. Click any text block to get word-by-word breakdowns, etymology, and contextual insight powered by AI.

4. Select & Explain (Any Webpage)
Select any text on any website. A popover instantly appears with AI-powered explanation. Works everywhere, not just PDFs.

5. 7 Theme Styles
Choose from 7 popover themes — Basic, Pro, Creative, Galaxy, Neon, Cyberpunk, Aurora.

━━━━━━━━━━━━━━━━━━━━

📄 SUPPORTED DOCUMENTS

✓ Scanned PDFs (paper documents digitized by scanner)
✓ Image-based PDFs (PDFs with embedded images instead of text layers)
✓ Image files (PNG, JPG)

Coming soon:
• Text-based PDFs (standard selectable-text PDFs)
• Multi-page processing
• EPUB / DOCX

━━━━━━━━━━━━━━━━━━━━

🔒 PRIVACY

• OCR runs 100% in your browser using local ONNX models
• No document is uploaded to any server for OCR
• AI translation/explanation uses API calls to LLM providers (SiliconFlow, Zhipu) — only the recognized text is sent, never the original document
• No analytics, no tracking, no account required

━━━━━━━━━━━━━━━━━━━━

⚙️ PERMISSIONS EXPLAINED

• Storage: Save your theme preferences and settings locally
• Host permissions (api.siliconflow.cn, open.bigmodel.cn): Required for AI translation and explanation API calls

No other permissions needed. No browsing history access. No tabs access.

━━━━━━━━━━━━━━━━━━━━

⚠️ ALPHA NOTICE

This is an early alpha release. Current limitations:
• Only the first page of a PDF is processed. Multi-page support is in development.
• Translation and explanation output is Simplified Chinese only.
• In-browser OCR works best with printed text. Handwriting is not supported.
• Very large files (>50 MB) may be slow due to browser memory limits.

We iterate fast. Feedback welcome.

━━━━━━━━━━━━━━━━━━━━

Website: https://scan-explain.vercel.app