Extract Text (OCR)
Extract text content from images using high-performance Optical Character Recognition.
The Extract Text (OCR) block uses advanced machine learning to "read" text within images. Whether you are scanning receipts, reading license plates, or extracting data from screenshots, this block converts visual information into editable text.
Core Concept
Think of this block as a digital translator that speaks "image" and "text." It looks at the shapes and patterns in an image, identifies characters, and reconstructs them into words and sentences.
Configuration
| Field | Required | Description |
|---|---|---|
| Input Image (Base64) | ✅ | The raw Base64 string of the image to scan. |
| Language | ✅ | The primary language of the text in the image. Selecting the correct language significantly improves accuracy. |
| Output Variable | ✅ | The name of the variable that will store the extracted text data object. |
Currently supports English, German, French, Spanish, Italian, Russian, Chinese (Simplified), and Japanese.
Error Handling
INPUT_MISSING— Triggered when no image data is provided for scanning.OCR_FAILED— Triggered if the OCR engine fails to process the image (e.g., due to extremely low resolution or corruption).
Example
- Input Image:
{captured_receipt_base64} - Language:
German - Output Variable:
receipt_data
Result: The variable {receipt_data} will contain an object like this:
{
"text": "Total: 45.00€\nDatum: 01.01.2024",
"confidence": 94.2,
"lines": ["Total: 45.00€", "Datum: 01.01.2024"]
}You can access the full text via {receipt_data.text} or individual lines via {receipt_data.lines}.