General questions about Vision OCR
Can I get logs of my operations with services?
Yes, you can request log records about your resources from Yandex Cloud services. For more information, see Data requests.
Which permissions are missing if an access error occurs when accessing the image recognition service?
Vision OCR returns the "Permission denied" error if the subject sending the request has no relevant access rights. To fix this error, assign the subject the editor
role or higher. For more information, see Access management in Vision OCR.
For which tasks can I use Vision OCR in Yandex Cloud?
You can use Vision OCR to:
- Recognize text
- Detect faces in an image
- Moderate images
- Determine image quality
- Encode files in Base64
For a detailed guide, see Step-by-step guides for Vision OCR.
Is Vision OCR suitable for recognizing documents (passport or driver's license)?
Yes, it is. Supported templates are listed in the Recognition models section. You can test other templates yourself.
license-plates
recognition model support?
Which languages does the The license-plates
recognition model supports only Russian (ru
) and English (en
). You need to specify them explicitly in your request, in the language_codes
parameter.
How do I recognize and present individual fields from a document rather than the whole text?
For this, select the line
model in the configuration: this model is better suited for recognizing individual lines. For a more detailed guide, see Recognizing a text line using the Vision API.
Why are element coordinates determined incorrectly?
Your image processor has incorrectly processed metadata from the exif
section. For more information, see Errors determining coordinates.
Do one of the following:
- Enable support for
exif
metadata in your image viewing tool. - Remove the
Orientation
attribute from the imageexif
section during transfer to the service or set it to0
.