Train OCR to perform better?

I’ve been testing the Free OCR with images to see if it will fit my needs and currently it comes the closest as compared to others that I’ve tried but it still isn’t collecting all of the information off of the image that I need. I’m wondering if there are things that I can do either with the image itself or with the engine settings that may help me to get a better result. Currently, this is all I get,

I’m looking to be able to get the names and the various stats in a table format that I can manipulate. The table part I can worry about later, just trying to get all the information to be read by the OCR at the moment. Any tips or tricks to make it work better would be appreciated.

  1. For table ocr use isTable = true in the OCR API call. The switch triggers the table scanning logic.

  2. Your main question is about OCR for single digit numbers. That is tricky. I have not yet found a good solution. OCR engine 2 and 5 are best for number ocr, but they still miss many single digit values.