How to get character coordinates in Computer Vision API?

Zhivko-2008 20

The Computer Vision Form Recognizer read (OCR) model response contains bounding boxes for words, but no character/glyph positions. Is there any way to get individual character positions?

romungi-MSFT 42,286 Reputation points Microsoft Employee

2023-02-01T11:16:00.9733333+00:00

Zhivko I noticed that your recent experience was not helpful, and to better improve our processes and learn from our customers, I'm eager to know what could have been done better.

Accepted answer

romungi-MSFT 42,286 Reputation points Microsoft Employee

2023-01-30T14:03:41.3733333+00:00

Zhivko The OCR and Read API can provide only the co-ordinates of each word from the lines that include all extracted words with their coordinates and confidence scores. Currently, there is no setting to get the individual character positions in the response.

-Please kindly accept the answer if the answer was helpful to support the community, thanks.
Please sign in to rate this answer.
romungi-MSFT 42,286 Reputation points Microsoft Employee

2023-01-31T05:36:48.17+00:00

@Zhivko I understand that the current limitation of the service did not help in this case and might have not answered your question. To add to my response above, there is a feedback portal for Azure services where you could log this requirement for the product team to review and assess the requirements as part of future releases. Please log your request here. I hope this helps!!
Sign in to comment

How to get character coordinates in Computer Vision API?

0 additional answers