OCR (Optical Character Recognition)

Guides

Extract text from images and documents using OCR.

What You’ll Build

An OCR pipeline to extract text from:

Scanned documents
Photos of text
Receipts and invoices
ID cards and forms
Screenshots

Prerequisites

A SeeMe.ai account (sign up)
Images containing text
(Optional) Python environment with seeme SDK installed

Supported Engines

SeeMe.ai supports multiple OCR backends:

PaddleOCR - Multi-language support, high accuracy
Tesseract - Open-source, widely used

Quick Start

from seeme import Client

client = Client()

## Get OCR model
models = client.get_models()
ocr_model = next(m for m in models if "ocr" in m.name.lower())

# Make prediction
result = client.predict(
    model_id=ocr_model.id,
    item="./document.png"
)

print(result.text)

Output Format

OCR returns structured results:

{
  "text": "Full extracted text...",
  "blocks": [
    {
      "text": "Line 1",
      "confidence": 0.98,
      "bbox": [10, 20, 200, 40]
    }
  ]
}