I'm curious as to what OCR library you are using? I've had issues getting 100% reads in the past on related projects, and I've love to hear how you deal with image de-skewing and OCR.