Use this skill whenever the user wants text extracted from images, photos, scans, screenshots, or scanned PDFs. Returns exact machine-readable strings with line-level text and optional bbox coordinates. Strong accuracy for CJK, small print, and handwritten text. Trigger terms: OCR, 文字识别, 图片转文字, 截图识字

PaddleOCR Text Recognition When to Use This Skill Use this skill for : Extract text from images (screenshots, photos, scans) Extract text from PDFs or document images when the goal is line/box level text Extract text from URLs or local files that point to images/PDFs Do not use for : Documents with tables, formulas, charts, or complex layouts — use Document Parsing instead Usage Basic OCR From URL: From local file: Common Options Output Format Important Notes Preprocessing options : By default, the API enables document preprocessing (unwarping and orientation classification). For flat, well oriented images (screenshots, properly scanned documents), you can disable preprocessing for faster results: Keep preprocessing enabled when: The input is a photo of a curved or folded document The document has significant perspective distortion Orientation is uncertain (rotated 90/180/270 degrees) Display complete results : Always show the full extracted content to users. Do not truncate with "..." unless content exceeds 10,000 characters. When multiple pages are processed, summarize if needed but provide complete results when explicitly requested. Handle errors gracefully : When the CLI returns an error, inform the user of the specific issue rather than silently failing or falling back to your own vision capabilities. Common errors: Authentication: PADDLEOCR ACCESS TOKEN invalid or missing Quota: API rate limit exceeded No content detected: Image may be blank or contain no text CLI Reference Run paddleocr api help for all options. For full documentation, see: [PaddleOCR Official Documentation](https://www.paddleocr.ai/latest/en/version3.x/inference deployment/serving/paddleocr official api/cli.html)

paddleocr-text-recognition