OCR using computer vision API for cropped images

arun prasath jp 1 Reputation point
2020-01-30T12:26:37.107+00:00

I have an issue regarding results generated from azure

I have one image which I want to OCR. (See full.jpg). This works fine with this image and all data is extracted. But when I crop a certain portion of this image and pass this as a new image (See crop.jpg), the OCR fails. The same part of this image in full.jpg is extracted fine,

Assistance would be appreciated

2466-crop.jpg2455-crop.jpg

Not Monitored
Not Monitored
Tag not monitored by Microsoft.
40,518 questions
0 comments No comments
{count} votes

2 answers

Sort by: Most helpful
  1. Michael Sharland 1 Reputation point
    2020-02-04T16:08:04.217+00:00

    I get similar results to you. I made my own crop of the image and using it will detect the word Quantity but not the numbers, but only if I set parameters to force the language to be English and set detectOrientation to false.

    0 comments No comments

  2. romungi-MSFT 48,036 Reputation points Microsoft Employee
    2020-02-14T04:34:54.353+00:00

    @arun prasath jp Improvements to the models to recognize single character images will be shipped in our future releases which should address this issue. We do not have an ETA for this release but we shall update the notes when we have more updates.

    0 comments No comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.