0

I am building a solver for KenKen, and for recognizing characters I am using easyocr. While testing the model, I hit a roadblock; it struggles with the division symbol. I tried to use the following code using easyocr==1.6.2:

import easyocr
reader = easyocr.Reader(['en'])
reader.readtext('1.png')

This was the image that was used as 1.png:

enter image description here

The output I am getting is sometimes an equals sign (=) in certain cases (different image configs) or sometimes a plus symbol (+), which would be okay if not for the fact that the plus sign is also used in KenKen.

vs07
  • 195
  • 8
  • I thought of manually adding the division symbol to the library, but that did not work. I've settled to instead utilize the probabilities that are given when a string of text is predicted; if the probability is low and it is a `+` or `=`, it is likely going to be a division symbol. – vs07 May 23 '23 at 23:22

0 Answers0