Hi,
I am trying to improve OCR with french plates, because I noticed that there is no "leu.france.exp0.box" file in the train-ocr/eu/input folder. So I guess, when given "fr" as a region, the OCR tries to match with German, Serbian, Dutch or Belgian fonts.
I gathered many plates pictures and i noticed that there is many different fonts for the french license plates. My question is : how different have to be the characters to be considered as part of different fonts ?
For example, i decided to process these two plates as part of two different fonts, because of the difference with the 7 and the 3. All the other characters from the fonts are almost the same. Can i mix them or not ?
Thanks in advance.

