Filled OMR Recognization Issue with Grid

Aspose.OMR Cloud Product Family
1

Hi,

I am able to generate the OMR without any issue.
After that i bubbled student id with a number where it is created with grid elements.
After scanning and uploading the omr sheet using, v5.0 api, it is returning unexpected results for student id and for answers too. I have created student id with 6 digits and api returning more digits than 6. And for answers, some answers are recognized with multiple answers where student bubbled filled 1 or nothing. Attaching all files for your investigation.

Need your help for below:

  1. Is there any standard guidelines for scanner while scanning?
  2. Pls let us know if anything is missing in template generation or .omr files.

Thanks in advance.
Recognization_issue.zip (405.9 KB)

2

Hello, @shashiakk

I have registered this issue as OMRCLOUD-221.

  1. Guidelines for scanning are next:
    1.1. Assert that all five reference points (black squares in the corners) are present in the image.
    1.2. Best quality of recognition is achieved when all four corners of the paper are equally distant from the center of the image (as little as possible angle or Z-axis rotation).
    1.3 Best quality of recognition is achieved when all four corners of the paper are as close to corners of the image as possible.
    1.4 Best quality of recognition is achieved when scanned image resolution is at least 1000x1000 pixels.
    1.5 Scans with more than 10’000x10’000 pixels may take more time to process.

I can confirm that, provided in attachment, scan satisfy all of the above criteria.

  1. Extra digits in “studentId” element is standard behaviour of Grid element.
    In short - it catches all marked digits, even if they are in the same row twice.
    CompositeGrid - catches one digit per row or replace whole result with “ERROR: Multiple marks per symbol” text.

  2. Regarding recognition issues and template matching, I replied in that thread