AksonOCR V2 API User Guide

AksonOCR Document AI - OCR Processor

AksonOCR V2 API provides advanced OCR (Optical Character Recognition) capabilities, powered by our advanced models: AksonOCR-preview, AksonOCR-handwriting, and AksonOCR-1.0. You can extract text and structured content from images and PDF documents with high accuracy and markdown output.

Key Features

• Extracts text content while maintaining document structure and hierarchy
• Preserves formatting like headers, paragraphs, lists, and tables
• Returns results in markdown format for easy parsing and rendering
• Handles complex layouts including multi-column text and mixed content
• Processes documents at scale with high accuracy
• Confidence scoring for each page and overall document (0-100 scale)
• Optional token-level confidence scores for granular accuracy assessment
• Supports multiple document formats:
  • image_url: PNG, JPEG/JPG, WEBP, and more
  • document_url: PDF and images (via URL or base64)
  • Base64-encoded images and PDFs

---

Endpoints

1. /api/v2/ocr (URL Upload)

Protected endpoint for OCR processing. Requires API key authentication.

Features

• API key authentication required
• Max 50 pages for PDFs
• Per-page credit charging (varies by model)
• API-based rate limiting

Supported Input Formats

• Image URLs (PNG, JPG, WEBP)
• Document URLs (PDF or images)
• Base64-encoded images
• Base64-encoded PDFs

Credits Costs vary by model (0.5-2 credits per page).

Example Requests

OCR with Image URL (with Token Confidence)

curl -X POST 'https://backend.aksonocr.com/api/v2/ocr' \
  -H 'X-API-Key: <YOUR_API_KEY>' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "AksonOCR-preview",
    "tokenConfidence": true,
    "document": {
      "type": "image_url",
      "image_url": "https://example.com/receipt.png"
    }
  }'

OCR with PDF URL

curl -X POST 'https://backend.aksonocr.com/api/v2/ocr' \
  -H 'X-API-Key: <YOUR_API_KEY>' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "AksonOCR-preview",
    "document": {
      "type": "document_url",
      "document_url": "https://arxiv.org/pdf/2201.04234.pdf",
      "document_name": "research_paper.pdf"
    },
    "pages": [0, 2, 5]
  }'

OCR with Base64 Image

curl -X POST 'https://backend.aksonocr.com/api/v2/ocr' \
  -H 'X-API-Key: <YOUR_API_KEY>' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "AksonOCR-preview",
    "document": {
      "type": "document_url",
      "document_url": "data:image/png;base64,iVBORw0KGgoAAAANS..."
    }
  }'

OCR with Base64 PDF

curl -X POST 'https://backend.aksonocr.com/api/v2/ocr' \
  -H 'X-API-Key: <YOUR_API_KEY>' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "AksonOCR-preview",
    "document": {
      "type": "document_url",
      "document_url": "data:application/pdf;base64,JVBERi0xLjQKJeLjz9MK..."
    }
  }'

Example Response

{
  "model": "AksonOCR-preview",
  "pages": [
    {
      "index": 0,
      "markdown": "# Invoice\n\n**Invoice Number:** INV-2025-001\n**Date:** 2025-10-29\n...",
      "confidence": 92,
      "tokens": [
        {
          "token": "Invoice",
          "confidence": 99.9
        },
        {
          "token": "Number",
          "confidence": 98.5
        }
      ]
    }
  ],
  "confidence": 92,
  "usage": {
    "pages_processed": 1
  }
}

2. /api/v2/upload (File Upload)

Upload endpoint for images and PDFs. Use this to upload files for OCR processing.

Features

• Accepts image files (PNG, JPG, WEBP) and PDFs
• Max file size: 10MB
• For PDFs, max 50 pages
• Returns extracted text in markdown format with confidence scores

Example Request (Image Upload with Token Confidence)

curl -X POST 'https://backend.aksonocr.com/api/v2/upload' \
  -H 'X-API-Key: <YOUR_API_KEY>' \
  -F 'file=@/path/to/image.png' \
  -F 'model=AksonOCR-preview' \
  -F 'tokenConfidence=true'

Example Request (PDF Upload)

curl -X POST 'https://backend.aksonocr.com/api/v2/upload' \
  -H 'X-API-Key: <YOUR_API_KEY>' \
  -F 'file=@/path/to/document.pdf' \
  -F 'model=AksonOCR-preview'

Example Response

{
  "model": "AksonOCR-preview",
  "pages": [
    {
      "index": 0,
      "markdown": "# Invoice\n\n**Invoice Number:** INV-2025-001\n**Date:** 2025-10-29\n...",
      "confidence": 92
    }
  ],
  "confidence": 92,
  "usage": {
    "pages_processed": 1
  }
}

---

Confidence Scoring

Each OCR response includes confidence scores to help you assess the reliability of the extracted text.

Response Fields

• pages[].confidence: Per-page confidence score (0-100)
• confidence: Overall confidence score (average of all pages)
• pages[].tokens: (Optional) Array of tokens with individual confidence scores. Include tokenConfidence: true in your request to receive this.

Example Multi-Page Response with Confidence

{
  "model": "AksonOCR-preview",
  "pages": [
    {
      "index": 0,
      "markdown": "# Page 1 Content\n\nExtracted text from page 1...",
      "confidence": 95
    },
    {
      "index": 1,
      "markdown": "# Page 2 Content\n\nExtracted text from page 2...",
      "confidence": 78
    }
  ],
  "confidence": 86,
  "usage": {
    "pages_processed": 2
  }
}

---

Notes

• Make sure your URLs are public and accessible by the API
• Only use supported formats for best results
• All results are returned in markdown for easy parsing
• For any issues, refer to error codes in the response

Error Codes

• MODEL_NOT_FOUND: Model parameter is invalid
• INVALID_INPUT: No file provided or invalid input
• PDF_CONVERSION_ERROR: Failed to process PDF file
• INSUFFICIENT_CREDITS: Not enough credits for processing
• RATE_LIMIT_EXCEEDED: Too many requests
• PROCESSING_ERROR: Internal server error

For further help, contact support or see the full API documentation.