eDocument OCR icon

eDocument OCR

Extract text from images using eDocument OCR API with Vietnamese language support

Overview

This node integrates with the eDocument OCR API to extract text from images. It supports two OCR modes: a fast mode optimized for printed and typed text, and an advanced mode capable of recognizing handwritten text (at slower speeds). The node can process images provided either as binary data from previous workflow nodes or by downloading them from a specified URL.

Common use cases include:

  • Automating data extraction from scanned documents, forms, or invoices.
  • Digitizing handwritten notes or signatures.
  • Extracting text content from images for further processing or analysis in workflows.

For example, you could use this node to automatically read typed text from PDF scans or to convert handwritten meeting notes into editable text.

Properties

Name Meaning
OCR Type Choose OCR processing type based on your content:
- Fast OCR (Typed Text Only): High-speed processing for printed documents, forms, and typed text.
- Advanced OCR (Handwritten Text): More comprehensive processing that can handle handwritten text (slower).
Input Type Source of the image to process:
- Binary File: Use image data from previous nodes.
- URL: Download image from URL.
Binary Property Name Name of the binary property containing the image data (required if Input Type is "Binary File").
Image URL URL of the image to process (required if Input Type is "URL").
Options Additional options:
- Return Confidence Score: Whether to return confidence score for extracted text.
- Return Coordinates: Whether to return coordinates of detected text regions.

Output

The node outputs JSON data containing the OCR results returned by the eDocument API. The output includes:

  • Extracted text content.
  • The original file name or URL used.
  • The OCR type selected ("fast" or "advanced").
  • The input type used ("binary" or "url").
  • Optionally, confidence scores for each detected text segment if enabled.
  • Optionally, coordinates of detected text regions if enabled.

If the input was binary data, the node does not output binary data itself but processes it internally to produce the JSON result.

Example output structure (simplified):

{
  "originalFileName": "image.png",
  "ocrType": "fast",
  "inputType": "binary",
  "text": "Extracted text here...",
  "confidenceScores": [...],         // Present if Return Confidence Score option enabled
  "coordinates": [...]               // Present if Return Coordinates option enabled
}

Dependencies

  • Requires an API key credential for the eDocument OCR service.
  • The node makes HTTP POST requests to the eDocument API endpoint at http://dms-be.eranin.com/office/ocr/document/v2/extract.
  • If using URL input type, the node downloads the image from the provided URL before sending it to the API.
  • The node expects the API key to be configured in the node credentials prior to execution.

Troubleshooting

  • No API key provided: The node will throw an error if the API key credential is missing or empty. Ensure you have configured the required API key credential.
  • No binary data found: When using binary input type, if the specified binary property name does not exist or contains no data, an error is thrown. Verify the binary property name matches the actual data.
  • Failed to download image from URL: If the image URL is invalid, unreachable, or returns an HTTP error, the node will fail with a descriptive message. Check the URL accessibility and correctness.
  • OCR processing failed: Common reasons include invalid API key (401 Unauthorized) or exceeding API quota limits (429 Too Many Requests). The error messages guide you to obtain a valid API key or purchase more credits.
  • General errors: Any unexpected errors during processing are caught and reported with the item index for easier debugging.

Links and References

Discussion