How to Run PDF OCR in n8n (Scanned Docs to Text)
Extract machine-readable text from scanned PDFs and route results to downstream systems.
OCR in n8n is ideal when you need searchable text from PDFs arriving via email, storage buckets, or forms.
Key OCR parameters
- pages for selective processing
- lang for language accuracy
- dpi to improve scan readability
- psm/oem for segmentation and engine behavior
Quality controls for OCR pipelines
Add field-level checks for key outputs such as totals, dates, and identifiers before writing results into your source-of-truth systems. If confidence is low or required fields are missing, route the document into a review queue instead of pushing incomplete data downstream.
Keep sample sets for recurring vendors or document formats and compare extraction quality over time. This gives you a practical baseline when adjusting DPI, language packs, and segmentation settings in automated n8n flows.
This baseline-driven tuning process is usually the fastest way to improve OCR consistency without overfitting your configuration to a single document template.
POST /api/v1/pdf/ocr/parse
{
"url": "https://.../scan.pdf",
"pages": "1-3",
"lang": "eng",
"dpi": 220,
"psm": 3,
"oem": 3
}Conclusion
Add confidence checks and fallback paths when extracted fields are empty or low quality. This protects downstream systems from bad data while preserving a review flow for edge-case scans.
Use OCR in n8n to convert unstructured scans into structured pipelines. Start with PDF OCR API.