Document Intelligence
Document Intelligence is a feature that extracts and organizes information from documents such as invoices, receipts, forms, contracts, and reports. It uses Optical Character Recognition (OCR) and document analysis techniques to process digital documents, scanned files, and images.
It identifies text, tables, key-value pairs, selection marks, and document layout elements, helping reduce manual data entry and simplify document processing. The extracted information can be used in business applications, databases, and automated workflows.
Azure Setup and Configuration
Document Intelligence can be accessed through Microsoft Azure AI Document Intelligence. Before configuring the service, ensure that the following prerequisites are met.
Prerequisites
a. An active Azure subscription.
b. Permission to create and access Azure resources.
c. An input document in a supported file format.
d. A document that complies with the service’s file size, page count, and image dimension limits.
The available features and processing limits depend on the selected model, API version, and Azure pricing tier.
Step 1: Create an Azure Document Intelligence Resource
1. Sign in to the Azure Portal.
2. Select Create a resource.
3. Search for Document Intelligence and select the service.
4. Select Create.
5. Configure the following details:
a. Subscription: Select your Azure subscription.
b. Resource Group: Select an existing resource group or create a new one.
c. Region: Select a supported Azure region.
d. Resource Name: Enter a unique name for the resource.
e. Pricing Tier: Select a suitable tier based on your usage requirements.
6. Select Review + create.
7. Review the configuration and select Create.
8. After deployment completes, select Go to resource.
Step 2: Retrieve the Endpoint URL and API Key
1. Open the created resource in the Azure Portal.
2. In the left navigation pane, select Keys and Endpoint.
3. Copy the URL displayed in the Endpoint field.
4. Copy KEY 1 or KEY 2.
Keep the endpoint URL and API key available for configuring the application or component that uses the service.
Step 3: Configure the Endpoint and API Key
Open the configuration of the application or component that connects to Azure Document Intelligence.
1. Enter the copied endpoint URL in the endpoint configuration field.
2. Enter the copied API key in the API key configuration field.
3. Save the configuration.
The exact parameter names and configuration location depend on the application or component being used.
Example endpoint URL: https://<your-resource-name>.cognitiveservices.azure.com/
Security Note: Treat API keys as sensitive credentials. Store them securely and do not expose them in public documentation, screenshots, source code, or logs. Ensure that the endpoint URL and API key belong to the same Azure resource.
Supported Formats and Limitations
Supported input formats and processing limits vary by model and API version. The following information describes commonly supported formats and the applicable limitations of the Azure AI Document Intelligence Layout model. Refer to the official Microsoft documentation for the latest service requirements.
Supported File Formats
| File Type | Supported Formats |
|---|---|
| Images | .jpg, .jpeg, .png, .bmp, .tif, .tiff, .heif |
| Word | .docx |
| Excel | .xlsx |
| PowerPoint | .pptx |
| HTML | .html |
File Size and Page Limits
| Restriction | Free Tier (F0) | Standard Tier (S0) |
|---|---|---|
| Maximum File Size | 4 MB | 500 MB |
| PDF and TIFF Page Processing | First 2 Pages | Up to 2,000 Pages |
These limits may vary depending on the model, API version, input format, and pricing tier. Check the current Azure service limits before processing documents.
Image Dimension Requirements
For image-based documents, ensure that the input image meets the following requirements:
a. Minimum image dimensions: 50 × 50 pixels.
b. Maximum image dimensions: 10,000 × 10,000 pixels.
c. Minimum text height: 12 pixels for a 1024 × 768 image, equivalent to approximately 8-point text at 150 DPI.
Additional Limitations
a. Image dimensions: Images must meet the service’s supported dimension requirements.
b. Password-protected PDFs: Remove password protection before submitting a PDF if the service cannot access it.
c. Document quality: Poor-quality scans, low-resolution images, unclear text, and complex layouts may affect extraction accuracy.
d. Office documents: For Word, Excel, and PowerPoint files, embedded or linked images are not supported by the Layout model. The model extracts supported embedded text.
e. Model capabilities: The Layout model extracts text, tables, selection marks, and document structure. Specialized document fields may require a dedicated prebuilt model.
f. Input requirements: Files must comply with the supported formats and applicable service limits.
Official Microsoft References
Refer to the following official Microsoft documentation for additional information.
1. Document Intelligence overview — General capabilities and service information.
2. Document layout analysis — Supported formats, Layout model capabilities, and input requirements.
3. Service quotas and limits — File size limits, page limits, quotas, and service restrictions.
4. Document processing models — Available models and their supported capabilities.
