How to format scans for optimal model performance (with examples) #1

Hi, thanks for the project. I have the model running, but the recognition rate is very poor. I think I’m not formatting the scans correctly.

What exact input format do the scans need?

Can you provide example scans that worked well in training or evaluation?

Are there preprocessing steps (normalization, filtering, tokenization) expected before feeding scans to the model?

Thank you!