The CloudOCR engine uses Microsoft's Computer Vision API to recognize handwritten, printed, and mixed text from documents, zones, and fields.
The CloudOCR engine specifically supports the following calls of Microsoft's Computer Vision API v2.0, 3.0, and 3.1.
-
[base
endpoint]/vision/v2.0/read/core/asyncBatchAnalyze
Note: This call supports only English text.
-
[base
endpoint]/vision/v3.[x]/read/analyze[?language=xx]
Note: The optional language parameter does not affect the manner in which BIC handles and stores the returned results.
For details on what is available, see Microsoft's Computer Vision documentation.
Prerequisites
-
A Microsoft Azure account
-
A Microsoft Azure Cognitive Services subscription.
This provides you with a subscription key and a base endpoint.
Sample Base Endpoint
https://brainware.cognitiveservices.azure.com/
Availability
CloudOCR is available for the following areas.
- Document recognition
- Zone recognition
- The anchor type Fit to Text
- The OCR optimization option Read field again
Notes
- BIC converts document pages to the appropriate image format.
- BIC handles the page
rendering and submits documents per page.
- For free tier Cognitive Services subscriptions, this effectively removes the two page per document limitation imposed by Microsoft.
- The maximum page size is 32 inches square at 300 DPI.