question

DannyGrady-1187 avatar image
0 Votes"
DannyGrady-1187 asked ·

OCR without assigning a language

I'm using the Computer Vision API (v3.1) OCR Post method.

I'm attempting to read some codes such as serial numbers. When I use the OCR Post method it attempts to assign a language to the image I've sent. Being that the code is not in any language it does not return any results. The irony is that the image is a white background, with OCR-specific font- it should be a piece of cake for any OCR method.

I wanted to use the OCR Post method because I would like to do a single method and avoid the callbacks of the Read method. Is there a way to use the OCR Post message and merely have the text sent back?

azure-computer-vision
10 |1000 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

1 Answer

ramr-msft avatar image
0 Votes"
ramr-msft answered ·

@DannyGrady-1187 Thanks for the question. The Computer Vision Read API is Azure's latest OCR technology that handles large images and multi-page documents as inputs and extracts printed text in Dutch, English, French, German, Italian, Portuguese, and Spanish. It also includes support for handwritten OCR in English, digits, and currency symbols from images and multi-page PDF documents. It's optimized to extract text from text-heavy images and multi-page PDF documents with mixed languages. It supports detecting both printed and handwritten text in the same image or document (for English only).

Please follow the doc for Read API.
https://docs.microsoft.com/en-us/azure/cognitive-services/computer-vision/concept-recognizing-text

· 2 ·
10 |1000 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

Hey @ramr-msft thanks for the reply.

If you noticed in my original post I said I wanted to avoid the callback of the Read method. Is there a way to utilize the OCR method such that it does not attempt to shoehorn this code into a language, and merely returns the text? I want one call, and the images I send (OCR-specific font, white background) will be ridiculously easy to parse.

That said- is the computer vision service overkill for this use case?

0 Votes 0 ·
ramr-msft avatar image ramr-msft DannyGrady-1187 ·

@DannyGrady-1187 Thanks for the details. The OCR API default value for language is "unk(AutoDetect)", then the service will auto detect the language of the text in the image. Can you please share the input image that you are trying to check. you can also use the Form Recognizer enables customers to extract data. The new version of Form Recognizer greatly expands language support, adds new capabilities like invoice line item extraction and natural reading order, and advances several core capabilities, with state-of-the-art deep learning technologies.


0 Votes 0 ·