Thought I'd reply on my own comment. Caffe has been used for handwriting, seems like OCR of typefaces would work just the same with a typeface dataset. https:

Thought I'd reply on my own comment. Caffe has been used for handwriting, seems like OCR of typefaces would work just the same with a typeface dataset. https://github.com/BVLC/caffe/tree/dev/examples/mnist https://github.com/BVLC/caffe/tree/dev/examples/mnist Here is the only other OCR example with caffe i could find: https://github.com/pannous/caffe-ocr https://github.com/pannous/caffe-ocr