Computer Vision API Frequently Asked Questions

Tip

If you can't find answers to your questions in this FAQ, ask the Computer Vision API community on StackOverflow or contact Help and Support on UserVoice

General Computer Vision questions

How can I increase the transactions-per-second (TPS) allowed by the service?

The free (S0) tier only allows 20 transaction per minute. Upgrade to the S1 tier to get up to 30 transactions per second. If you're seeing the error code 429 and the "Too many requests" error message, submit an Azure support ticket to raise your TPS to 50 or higher with a brief business justification. Computer Vision pricing.

The service is throwing an error because my image file is too large. How can I work around this?

The file size limit for most Computer Vision features is 4 MB, but the client library SDKs can handle files up to 6 MB. For Optical Character Recognition (OCR) that handles multi-page documents, the maximum file size is 50 MB. For more information, see the Image Analysis inputs limits and OCR input limits.

How can I process multi-page documents with OCR in a single call?

Optical Character Recognition, specifically the Read operation, supports multi-page documents as the API input. If you call the API with a 10-page document, you'll be billed for 10 pages, with each page counted as a billable transaction. If you have the free (S0) tier, it can only process two pages at a time.

Can I send multiple images in a single API call to the Computer Vision service?

This function isn't currently available.

How many languages are supported for Computer Vision services?

See the Language support page for the list of languages covered by Image Analysis and OCR.

OCR service questions

How can I process multi-page documents with OCR in a single call?

Optical Character Recognition, specifically the Read operation, supports multi-page documents as the API input. If you call the API with a 10-page document, you'll be billed for 10 pages, with each page counted as a billable transaction. Note that if you have the free (S0) tier, it can only process two pages at a time.

Can I deploy the OCR (Read) capability on-premises?

Yes, the OCR (Read) cloud API is also available as a Docker container for on-premises deployment. Learn how to deploy the OCR containers.

Image Analysis service questions

Can I train Computer Vision API to use custom tags? For example, I would like to feed in pictures of cat breeds to 'train' the AI, then receive the breed value on an AI request.

This function is currently not available. You can use Custom Vision to train a model to detect user-defined visual features.