ocr

You can extract text from scanned PDF files using OCR to convert images into machine-readable text. This skill processes legal evidence and filings using PaddleOCR and pytesseract, outputting searchable PDFs and .docx reports for case files.

Yes, you can batch process a directory of scanned images. The skill supports both single-file and directory batch processing for common image formats and scanned PDFs, generating per-page confidence scores and structured JSON outputs.

OCR works on low quality scans by applying image preprocessing techniques like deskewing, contrast enhancement, and noise reduction. These steps improve text extraction accuracy on degraded scans and faxes before running the OCR engine.

The best way to make archived legal documents searchable is using language-aware OCR to extract text and export a searchable PDF layer. This creates machine-readable text from scanned filings, enabling search and summarization without manual retyping.

You can export OCR results to plain text, structured JSON, searchable PDF layers, and .docx formats. These outputs include human-readable extraction reports with confidence scores for review and integration into legal case files.

You need PyMuPDF or poppler-utils to process scanned PDFs, while optional Node.js tooling provides additional functionality. The core OCR pipeline relies on PaddleOCR, pytesseract, and Pillow, with a pytesseract fallback for lighter environments.