Add recognition of stacked characters (enhancement)

[ssmelov]

I think it might be good if we add a recognition of stacked characters on a license plates.

[matthill]

It's possible. We would just add a training dataset (for each state font) with those characters. However, the additional characters could lower overall accuracy (for example, with the extra font it may see an "8" as a "G/A" whereas without those extra characters it would match.

[SmrityB]

Is there any way figured out for the detection of stacked characters?