Nigel Kendrick wrote:
FuzzyOcr is proving to be useful but it does seem to be a bit too 'Fuzzy' at
times...
First of all, try lowering the focr_threshold to 0.25 or even lower
Secondly, add custom thresholds for the rules that misfire
For example change the line with 'best' to
best::0.2
So that it only fits if 'best' is found with a fuzz *below* 0.2

Matt

Reply via email to