n8n-nodes-deep-ocr

n8n-nodes-deep-ocr

N8N Community Node for the Deep-OCR Service — AI-powered OCR for invoices, receipts, contracts, IDs, bank statements and more — returns structured JSON ready to use.

What it does

You hand it a document, it hands you back clean JSON. No regex, no template matching, no per-vendor parser.

• Invoices → vendor, line items, tax breakdown, totals, IBAN, due date
• Receipts → merchant, items with per-item tax, payment method
• Bank statements → account holder, balances, full transaction list
• Contracts → parties, dates, obligations, governing law, signatures
• IDs / passports → personal data + MRZ lines
• Payslips, delivery notes, purchase orders, handwriting → and more

Example: invoice PDF → structured JSON

{
  "vendor": { "name": "ACME GmbH", "vat_id": "DE123456789" },
  "invoice_number": "RE-2024-00842",
  "invoice_date": "2024-03-15",
  "due_date": "2024-04-14",
  "line_items": [
    { "description": "Consulting", "quantity": 8, "unit_price": 150.00, "total": 1200.00 }
  ],
  "subtotal": 1200.00,
  "tax_amount": 228.00,
  "total": 1428.00,
  "iban": "DE89 3704 0044 0532 0130 00"
}

(Every response also includes document_type and a metadata object — see Output Examples below for the full shape.)

Drop it into any workflow that needs accounting, expense tracking, KYC, or document classification.

Features

• Structured Data Extraction: Receive a structured JSON object with the fields relevant to the document type
• Document Types & Auto Detection: Invoice, Receipt, Contract, Delivery Note, Bank Statement, Payslip, Purchase Order, ID Document, Handwriting, Full-Text, Generic, and automatic type detection (auto)
• Multiple Format Support: PDF, PNG, JPG, JPEG, WebP, TIFF (up to 100MB)
• Secure Authentication: API key stored securely using n8n credentials

Installation

Community Nodes (Recommended)

1. Go to Settings → Community Nodes
2. Click Install a community node
3. Enter n8n-nodes-deep-ocr
4. Click Install

Manual Installation

npm install n8n-nodes-deep-ocr

Configuration

Setting Up Credentials

1. Get your API key from Deep-OCR Dashboard
2. In n8n, go to Credentials → Add Credential
3. Search for "Deep-OCR API"
4. Enter your API key and save

Usage

1. Add the Deep-OCR node to your workflow
2. Connect a node that provides binary data (e.g., Read Binary File, HTTP Request)
3. Configure:
   • Binary Property: Name of the binary property containing your document (default: data)
   • Document Type: Select the type that matches your document
4. Execute — the node outputs a JSON object with the extracted fields

Document Types

Type	Description
generic	Flexible extraction for any document — adapts structure to the content. Cheap default; pick a specific type below when you know it
invoice	Vendor, customer, line items, tax breakdown, totals, IBAN, payment terms
receipt	Merchant, items, per-item tax rate, tax breakdown, totals, payment method
contract	Parties (role/name/address), dates, obligations, governing law, signatures
delivery_note	Sender, recipient, items (ordered vs. delivered), tracking, partial delivery info
bank_statement	Account holder, IBAN/BIC, opening/closing balance, full transaction list
payslip	Employer, employee (incl. IBAN), earnings, deductions, gross/net salary
purchase_order	Buyer, supplier, delivery address, line items, totals
id_document	Type, document number, personal data, MRZ lines
handwriting	Full transcription with unclear-word markers, confidence rating, language
fulltext	Transcribe the entire document as Markdown — billed at 10× a structured extraction
auto	Classify the document first and then extract — costs more than picking a specific type. Use only when the type is genuinely unknown

Output Examples

Invoice

{
  "vendor": { "name": "ACME GmbH", "address": "Musterstr. 1, 10115 Berlin", "vat_id": "DE123456789" },
  "invoice_number": "RE-2024-00842",
  "invoice_date": "2024-03-15",
  "due_date": "2024-04-14",
  "line_items": [
    { "description": "Consulting Services", "quantity": 8, "unit_price": 150.00, "total": 1200.00 }
  ],
  "subtotal": 1200.00,
  "tax_rate": 19,
  "tax_amount": 228.00,
  "total": 1428.00,
  "iban": "DE89 3704 0044 0532 0130 00",
  "document_type": "invoice",
  "metadata": { "pages": 1 }
}

Receipt

{
  "merchant": { "name": "REWE City", "address": "Friedrichstr. 12, 10117 Berlin" },
  "date": "2024-03-22",
  "items": [
    { "description": "Bioland Äpfel 1kg", "quantity": 1, "price": 2.49 },
    { "description": "Mineralwasser 6x1.5L", "quantity": 1, "price": 3.99 }
  ],
  "subtotal": 6.48,
  "tax_breakdown": [{ "rate": 7, "amount": 0.42 }],
  "total": 6.90,
  "payment_method": "EC-Karte",
  "document_type": "receipt",
  "metadata": { "pages": 1 }
}

Example Workflow

{
  "nodes": [
    {
      "name": "Read Invoice PDF",
      "type": "n8n-nodes-base.readBinaryFile",
      "parameters": {
        "filePath": "/path/to/invoice.pdf"
      }
    },
    {
      "name": "Extract Invoice Data",
      "type": "n8n-nodes-deep-ocr.deepOcr",
      "parameters": {
        "binaryPropertyName": "data",
        "documentType": "invoice"
      }
    }
  ]
}

Supported File Types

Format	MIME Type	Max Size
PDF	application/pdf	100MB
PNG	image/png	100MB
JPG/JPEG	image/jpeg	100MB
WebP	image/webp	100MB
TIFF	image/tiff	100MB

Troubleshooting

"Unsupported file type" error

The node only accepts PDF, PNG, JPG, JPEG, WebP, and TIFF files. Make sure the upstream node sets the correct MIME type. If you're using HTTP Request to download a file, check that the response has a proper Content-Type header.

"File size exceeds maximum allowed size"

The Deep-OCR API accepts files up to 100MB. For larger PDFs, consider compressing them first or splitting them into individual pages.

"Connection tested successfully" but processing fails

Your API key is valid, but the key may not have enough quota remaining. Check your usage in the Deep-OCR Dashboard.

Node not appearing in n8n's node picker

In n8n 2.x, unverified community nodes are hidden from the picker search. Install the node via Settings → Community Nodes, then add it to your workflow by importing the following JSON snippet:

{
  "type": "n8n-nodes-deep-ocr.deepOcr",
  "parameters": { "binaryPropertyName": "data", "documentType": "invoice" }
}

Binary property not found

The Binary Property field must match the property name set by the previous node. The default is data (used by Read Binary File, HTTP Request, etc.). Check the output of the previous node to confirm the property name.

Development

Prerequisites

• Node.js 22+
• pnpm 11+

Setup

# Clone the repository
git clone https://github.com/Heey-Global/deep-ocr-n8n.git
cd deep-ocr-n8n

# Install dependencies
pnpm install

# Build
pnpm build

# Run tests
pnpm test

# Run linter
pnpm lint

Project Structure

src/
├── credentials/
│   └── DeepOcrApi.credentials.ts    # API key credential type
├── nodes/
│   └── DeepOcr/
│       ├── DeepOcr.node.ts          # Main node implementation
│       └── deepocr.svg              # Node icon
├── utils/
│   └── errors.ts                    # Error handling utilities
└── index.ts                         # Package entry point

License

MIT

Contributing

Contributions are welcome! Please submit pull requests to our repository.

Support

• GitHub Issues
• Deep-OCR Documentation