Table, Chart & Form Extractor

Extract structured data from tables, charts, and forms in document images using AI vision analysis.

Overview

This Actor analyzes document images and automatically extracts:

• Tables — Column headers, rows, and captions as structured arrays
• Charts — Chart type, title, axis labels, data points, and summaries
• Forms — Field names, values, and field types (text, number, date, checkbox, select)

Perfect for processing invoices, financial reports, surveys, datasheets, data visualizations, and application forms.

Features

• Local AI Processing — Uses Qwen2.5-VL vision model running locally (no cloud APIs)
• Batch Processing — Process up to 20 images per run
• Structured Output — JSON format with confidence scores
• Error Handling — Graceful failures with detailed error codes
• Production Ready — Timeout protection, validation, retry logic

Input

Provide either a single image or array of images:

Single Image

{
  "imageUrl": "https://example.com/document.png"
}

Batch Processing

{
  "imageUrls": [
    "https://example.com/doc1.png",
    "https://example.com/doc2.jpg",
    "https://example.com/doc3.webp"
  ],
  "maxUrls": 20
}

Output

Each result contains:

{
  "imageUrl": "https://example.com/document.png",
  "status": "success",
  "contentType": "table|chart|form|mixed|unknown",
  "tables": [
    {
      "headers": ["Column 1", "Column 2", "Column 3"],
      "rows": [
        ["Row1Col1", "Row1Col2", "Row1Col3"],
        ["Row2Col1", "Row2Col2", "Row2Col3"]
      ],
      "caption": "Table title or description"
    }
  ],
  "charts": [
    {
      "chartType": "bar|line|pie|scatter|other",
      "title": "Chart Title",
      "xAxis": "X-axis Label",
      "yAxis": "Y-axis Label",
      "dataPoints": [
        {"label": "Q1", "value": "100"},
        {"label": "Q2", "value": "150"}
      ],
      "summary": "What this chart shows"
    }
  ],
  "forms": [
    {
      "fields": [
        {"fieldName": "Name", "fieldValue": "John Doe", "fieldType": "text"},
        {"fieldName": "Age", "fieldValue": "30", "fieldType": "number"},
        {"fieldName": "Agreement", "fieldValue": "checked", "fieldType": "checkbox"}
      ]
    }
  ],
  "rawText": "All text content from the image",
  "confidence": 0.95,
  "model": "Qwen2.5-VL",
  "processingTimeMs": 1500
}

Error Response

{
  "imageUrl": "https://example.com/document.png",
  "status": "error",
  "error": "AI service request timeout (30000ms)",
  "code": "TIMEOUT|INVALID_URL|INVALID_INPUT|API_ERROR|PROCESSING_ERROR",
  "processingTimeMs": 30100
}

Pricing

$0.15 per extraction

Charged only on successful extraction. Errors are not charged.

Quality Assurance

Best Results With

• High-resolution images (≥ 300 DPI)
• Clear, well-lit documents
• Standard table layouts
• Legible charts and forms

Limitations

• Complex nested tables may show reduced accuracy
• Handwritten forms may be less accurate than printed
• Artistic or stylized charts may require manual verification
• Images with overlapping elements may yield incomplete extraction

Legal Notice

Document Source Responsibility (IMPORTANT)

• You are solely responsible for ensuring you have the right to process uploaded documents.
• Permitted sources:
  • Documents you own or created
  • Public government filings (SEC EDGAR, USPTO patents, etc.)
  • Open access publications (CC-BY, CC0 licensed)
  • Documents you are licensed to analyze
• Prohibited sources:
  • Confidential or trade secret documents belonging to third parties
  • Documents containing insider information (pre-publication financial data)
  • Medical records or health insurance data (HIPAA protected)
  • Credit reports or financial records with personal data (FCRA/GDPR protected)

Copyright

• Extracted data from copyrighted documents may not be redistributed or sold commercially without the copyright holder's permission.
• Government public filings (SEC, USPTO) are generally in the public domain.
• Corporate annual reports and research publications are copyrighted; extraction for personal analysis may qualify as fair use, but commercial redistribution likely does not.

Accuracy & Liability

• Extraction accuracy depends on image quality and document complexity.
• All extracted data should be verified before use in financial, legal, or business decisions.
• This tool does not provide financial analysis, investment advice, or legal counsel.
• The developer assumes no liability for decisions made based on extracted data.

Data Processing

• Documents are processed on our local AI server and immediately discarded.
• No documents are stored, cached, or shared with third parties.
• We do not retain extracted data beyond delivering results to your Apify dataset.

Examples

Extract Invoice Table

curl -X POST https://api.apify.com/v2/acts/YOUR_ACTOR_ID/runs \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "imageUrl": "https://example.com/invoice.png"
  }'

Batch Extract Multiple Reports

curl -X POST https://api.apify.com/v2/acts/YOUR_ACTOR_ID/runs \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "imageUrls": [
      "https://example.com/report1.jpg",
      "https://example.com/report2.jpg",
      "https://example.com/survey.png"
    ],
    "maxUrls": 20
  }'

Support

For issues or feature requests, contact support.

License

See LICENSE file for terms.