Open jonatino opened 1 year ago
You could try some of the tricks from:
https://tesseract-ocr.github.io/tessdoc/ImproveQuality.html
What worked best for me was applying a mask or thresholding (https://docs.opencv.org/4.x/d7/d4d/tutorial_py_thresholding.html) and rescaling
I'm trying to read current hp levels from the rs client but when I try to use any of the models in your repo I get close, but incorrect values.
Image being ocr'ed:
Text output from ocr: osrs:
Hp is '53'
osrs2:Hp is '323'
osrs3:Hp is '3'
osrs-ft:Hp is '7'
Any advice for getting a correct reading?