I tried using Tesseract to OCR just some numbers on an almost plain background, and it failed around 2-3% of the time. Which made the whole thing useless, becau

I tried using Tesseract to OCR just some numbers on an almost plain background, and it failed around 2-3% of the time. Which made the whole thing useless, because I needed 100% correctness.