Computer vision OCR size matters ?

Question

I use azure computer vision 4.0 to recognize bib numbers in running race photos. Why do I get different results when I send a photo of 5000x3000 compared to when I send the same at 50% of that size? Sometimes the results are better on the larger one and sometimes they are better on the smaller one. In both cases, they are JPGs of the same quality.

Is there something I can do to increase success?

Answer 1

Hello @Courses Thematiques , Thanks for using Microsoft Q&A Platform.

These are the following input requirements for the OCR:

• Supported file formats: JPEG, PNG, BMP, PDF, and TIFF
• For PDF and TIFF files, up to 2,000 pages (only the first two pages for the free tier) are processed.
• The file size of images must be less than 500 MB (4 MB for the free tier) and dimensions at least 50 x 50 pixels and at most 10,000 x 10,000 pixels. PDF files don't have a size limit.
• The minimum height of the text to be extracted is 12 pixels for a 1024 x 768 image. This corresponds to about 8 font point text at 150 DPI.

If the character size is too small, which hits the limitation of the current OCR system you may face issues.

Please go through this documentation that describes the best practices to improve OCR system performance.

If the document intelligence Read model aligns with your use case, you can check with that.

Also, you can try image processing steps mentioned by @Vahid Ghafarpour .

I hope this helps.

Regards,
Vasavi

-Please kindly accept the answer and vote 'yes' if you feel helpful to support the community, thanks.