Text this: A survey of mono- and multi-lingual character recognition using deep and shallow architectures